DESIGN AND MICROFABRICATION OF A MOLDED POLYCARBONATE
CONTINUOUS FLOW POLYMERASE CHAIN REACTION DEVICE
A Thesis
Submitted to the Graduate Faculty of the
Louisiana State University and
Agricultural and Mechanical College
in partial fulfillment of the
requirements for the degree of
Master of Science in Mechanical Engineering
in
The Department of Mechanical Engineering
by
Michael Mitchell
B.S., Louisiana State University, 2000
December, 2002
ACKNOWLEDGEMENTS
First let me give GOD the praise and accept the glory of having accomplished an
educational goal set for myself four and half years ago. I would like to take this
opportunity to express my gratitude to my family, friends and colleagues who have
always been a source of strength and encouragement. The moment I had been waiting
for has finally arrived and I am excited to start a new phase in my life.
I am grateful to my major professor Dr. Michael Murphy for his support,
guidance and for his extreme levels of patience throughout my graduate studies at LSU.
I am still impressed by his ideas and his analytical approach to problems. I am
extremely appreciative of his guidance both, from a technical and professional front.
I thank my committee members Dr. Dimitris E. Nikitopoulos, Dr. Srinath V.
Ekad and Dr. Steven A. Soper for their comments and suggestions. I thank Dr. X. liu for
hip help with the microfabrication and experimental testing of the device. Dr. Srinath V.
Ekad for his valuable suggestions, ideas and outlook on the various heat transfer
problems arise during the design of this PCR device. I would like to thank, Dr. Dimitris
E. Nikitopoulos and Y. Bejet for their significant contribution toward the fluid
dynamics of this project. Thanks to Dr. Steven A. Soper for the all the journal article on
PCR device that he had provided, that I used as, references during my research. I thank
the entire staff at CAMD for given the privilege to use the facility during the fabrication
of my project.
This work was supported by the Bioengineering Research Partnership through the
National Human Genome Research Institute (NHGRI) and the National Cancer Institute
(NCI) of the National Institutes of Health (NIH).
ii
TABLE OF CONTENTS
ACKNOWLEDGEMENTS
……………………………………………….…...
ii
LIST OF TABLES
……………………………………………………
vi
LIST OF FIGURES
……………………………………………………
vii
…………………………………………………….……
xi
…………………………………………………………….…….
xv
………………………………….……….
CHAPTER 1. INTRODUCTION
1.1 Requirements for PCR Devices …………………………………………...
1.2 Conventional PCR Device …………… .…………………………….…..
1.3 Contribution to PCR Device Production
…………………………………
1.4 Thesis Outline ……………………………………………………………..
1
2
3
3
4
CHAPTER 2. BACKGROUND ………….…………………….……………..
2.1 Applications
…………………………………………………….………
2.2 Genetics
………………………………….……………………………
2.3 Literature Review of PCR devices
……….…………………………….
6
7
8
9
CHAPTER 3. DESIGN OF TEST DEVICE AND FINITE FLEMENT …
(THERMAL) ANALYSIS
3.1 Introduction to The Continuous Flow Straight Channel PCR Device …
3.2 Material Selection ……………………………………….………………...
3.3 Device Configuration
……….………………………….
3.4 Finite Element Analysis (ANSYS /FLOTRAN)
………………………
3.5 Steady-State Thermal
………...………….……….…………………….
3.6 Analysis of Air Pocket ………...………….……….…………………….
3.7 Problem Description
………...………….……….…………………….
3.8 Thermal Radiation and Conduction
…………………
3.9 Finite Element Analysis for Preliminary Simulations
…………………
3.10 Heater Positioning and Location
.………………………………………
3.11 Fluid Analysis of Test device
………………………………………
3.12 Coupled Analysis of The Test Device …………………………………..
3.13 FORTRAN (CFD) Element Type
…………………………………….
3.14 Problem Configuration
.….…………………………………………
3.15 Mesh Sizes
……………………………………………..
3.16 Finite Element Simulation Results
…..…………………………………
3.17 Conclusion of Finite Element Simulation
……………………………….
13
LIST OF SYMBOLS
ABSTRACT
…
CHAPTER 4. FABRICATION OF THE MOLD INSERT BOTH PCR
DEVICES
4.1 Introduction to (LIGA) Mold Insert
…………………………………….
iii
13
14
14
16
17
17
19
22
24
26
30
34
34
36
38
40
48
50
50
4.2 Background of LIGA Fabrications
……………………………...……
4.3 Optical Lithography
………….………..…………….………………
4.4 Optical Mask Design
…………..……………...……………….………
4.5 X-Ray Lithography
..………………………………….……………
4.6 X-Ray Mask Fabrication Techniques …………….….………….………..
4.7 Spin Coating and Pre-Bake of Resist
………….………………………
4.8 UV Exposure Post-Bake and Development of Resist
…………………
4.9 Graphite Substrate Preparations and Gold Plating
…………
4.10 Preparation of Stainless Steel Mold Insert Substrate ….………...………
4.11 PMMA Bonging
……………………………….…………………….
50
51
52
54
55
56
56
57
58
59
4.12 X-Ray Exposure of PMMA
4.13 PMMA Resist Development
…………..………………………………...
……………………...…...…….…………...
61
62
4.14 Nickel Electrodeposition
4.15 Lapping and Polishing
..……………………………………………
…….……….………………………………
63
65
CHAPTER 5. THERMAL SYSTEM DYNAMICS AND CONTROL ……
5.1 Introduction to System Dynamics of The PCR Device
……..…………
5.2 Pseudo Bond Graph Modeling for Thermal Systems
…………………
5.3 Control Systems Background
..……………………………………….
5.4 Stability
..……………………………………………
5.5 Heater Power Requirement
…………………………………………….
5.6 PID Controller Parameters
……………….……………………………
5.7 Heater Selection
...……………..…………………………….…………
5.8 Introduction to Heater Design
…………………………………………
5.9 Resistive Heating
………….……………………………….………
5.10 Design Specifications and Mask Layout ………...……………………….
5.11 Direct Electrodeposition on Polycarbonate
……………………………
5.12 Electrodeposition of Ni-Cr Alloy
5.13 Plating Experiments
…………………………………………………..
69
69
69
73
75
75
75
78
79
79
82
86
87
88
CHAPTER 6. FABRICATION ASSEMBLY AND EXPERIMENT ………
SET UP OF TEST DEVICE
6.1 Introduction to Fabrication and Assembly of Test Device
…..…………
6.2 Polycarbonate Molding
………………………………………..……….
6.3 Polycarbonate Bonding
…………………………………………………
6.4 Fluid Interconnects
…….………………………………………..…
6.5 Assembly of Test device
…………………………………………………
6.6 Control System Circuit Assemble
……………………
6.7 Experimental Set Up and Results
……………………
92
92
92
95
96
96
97
100
CHAPTER 7: DISCUSSIONS AND CONCLUSION ……………………..
7.1 Recommendations for Future Work
………………………………...
102
104
……………………………….………………...……………..
106
REFERENCES
iv
APPENDIX A: A SCHEMATIC OF THE FABRICATION
PROCEDURES USED TO MAKE THE GRAPHITE X-RAY MASK
…..
111
…..
112
…..
113
………………………….………………..…………….
114
APPENDIX B: A SCHEMATIC OF THE FABRICATION
PROCEDURES USED TO MAKE THE TWO MOLD INSERTS
USING THE X-RAY MASK
APPENDIX C: DIMENSION DRAWING OF THE TEST DEVICE
OPTICAL MASK DESIGN
VITA
v
LIST OF TABLES
Table 2.1
Cited works of prior micro PCR devices
10
Table 3.1
The design parameters and specifications
31
Table 3.2
The different element detentions of the three mesh
configurations
The temperature transition length between each zone
42
Table 3.4
The variation in the input heat fluxes for the different
velocities to obtain the applied temperature constraint
44
Table 4.1
Composition by mass of MMA bonding glue
60
Table 4.2
Composition by volume of the GG Developer formulated
at the Institute for Microstructure Technology at the
Forschungszentrum, Karlsruhe. This solvent is used to
dissolve PMMA that were exposed to X-rays.
Table 4.3
Composition by mass of the nickel sulfamate solution
64
Table 5.1
Numerical values for the resistors and capacitors for the
lumped parameter model
70
Table 5.2
The effects of each of the controller gain on a closed-loop
system
76
Table 5.3
The comparison of the desire specifications of the
electrical heaters to what were commercially available
78
Table 5.4
Electrical Properties of Ni-Chrome alloy
83
Table 3.3
Table 5.5
39
62
The bath composition for Ni-Cr alloy plating
88
vi
LIST OF FIGURES
Figure 1.1
The desired temperature profile for a single PCR cycle
3
Figure 3.1
A cross-sectional schematic view of the straight channel
device (not to scale)
15
Figure 3.2
Natural-convection flow regimes for a layer of fluid between
tow horizontal isothermal plates.
18
Figure 3.3
The model of the rectangular cavity simulated for the air
pocket
20
Figure 3.4
The contour of the temperature distribution across the heated
cavity
21
Figure 3.5
The contour of the pressure increase within the heated cavity
21
Figure 3.6
Two parallel resistors (radiation and conduction) across the air
pocket
23
Figure 3.7
Figure 3.8
Figure 3.9
Figure 3.10
The graph showing the comparison of the temperature
distribution across the air pocket, when simulated with
radiation and conduction and conduction only with the
equavalent thermal conductivity.
The cross-section of simulated straight section of the device,
which contained two constant temperature zones and one
transition zone
ANSYS simulated temperature contour obtained for the
device, each colors represent approximately a 100C
temperature bond where different colors showing the
temperature variation. (a) The device with fins and (b) The
device without fins
ANSYS simulated temperature contour obtained for the
device, where different colors showing the temperature
variation with a 100C temperature bond. (a) The device with
heaters one top and bottom and (b) The device with heaters on
top only.
Figure 3.11
Cross section of the rectangular micro duct
Figure 3.12
Three-dimensional velocity profile of the pressure driven flow
in the PCR channel
vii
25
27
29
29
32
33
This shows two cross sectional views of the model one in the
xy plane and the other in the yz plane
37
Figure 3.14
This is a three-dimensional view of the simulated section of
which two views are shown in Figure 3.13
38
Figure 3.15
The cross sections of the right end for each of the three
different mesh sizes of the fluid channel and its surrounding
solid region
Figure 3.16
The cross section of the three different mesh sizes for each
temperature transition zones
Figure 3.17 (a)
The comparison of the analytical velocity profile to the
simulated result obtained from ANSYS
40
Figure 3.17 (b)
An enlargement of the comparison of the analytical and
simulated velocity profiles
41
Figure 3.13
38
39
Figure 3.18
The comparison of the temperature distribution alone the
channel for the model with heaters on one side and with
heaters on both sides
Figure 3.19 (a)
The temperature contour across the device with heaters one
43
Figure 3.19 (b)
The temperature contour across the device with heaters on both
sides
43
Figure 3.20
The variation of temperature distribution along the
microfluidic channels, showing how increasing velocity affect
the temperature profile, with the design temperature constraint
Figure 3.21
Residence times ∆t72 , ∆t95 and ∆t55 , are the time the fluid spent
in the 720C, 950C and 550C zones, respectively, while ∆t72-95 ,
∆t95-55 and ∆t55-72 , are the times the fluid spent in transition
from 720C to 950C, 950C to 550C and 550C to 720C.
Figure 3.22
The ratio of the residence times of the fluid in the 720C, 950C
and 550C zones as function of mean velocity.
Figure 3.23
The ratio of the transition times for the fluid from 720C to
950C, 950C to 550C and 550C to 720C in the 720C, 950C and
550C as function of mean velocity
viii
42
44
46
47
47
Figure 3.24
The comparison between the temperature at the channel wall
and center for a velocity of 4 mm/s.
48
Figure 4.1
Optical mask pattern for the PCR straight channel designs
53
Figure 4.2
Optical mask pattern for the PCR spiral device conjuration
designs by Yannick Bejat
54
Figure 4.3
Fabricated X-ray mask for PMMA resist exposure.
58
Figure 4.4
SEM of straight channel mold insert
66
Figure 4.5
Actual photo of the straight channel mold insert
67
Figure 4.6
SEM of spiral channel mold insert
67
Figure 4.7
SEM of spiral channel mold insert
68
Figure 5.1
Equivalent electrical circuit for one dimensional heat flow for
each zone
72
Figure 5.2
Bond graph of the system
72
Figure 5.3
Ideal system with an open loop control
73
Figure 5.4
Typical system under closed loop control
74
Figure 5.5
The response of the system for a step input power of 1.415W
and 2.5W
74
Figure 5.6
The block diagram used in SIMULINK to determine the
correct gains for the simulation
76
Figure 5.7
The transient temperature response for each zone with PID
control
Figure 5.8
The mask pattern for the Ni-Cr heater with 24 turns all units
are in µm
Figure 5.9
The variation of heating element (Ni-Cr) resistance with
voltage for a power of 4W
ix
77
84
84
The temperature increase with an increase in current flow
through the heating element Comparison between the Lumped
parameter and finite element results simulated results
85
Figure 5.11
The heat flux from the heating element to the surrounding air,
which is equivalent to the heaters’ Power density.
86
Figure 5.12
The Hull Cell bath in Cartesian coordinate
89
Figure 5.13
Hull Cell current distribution along the working electrode with
average current of 12.5mA/cm2
90
Nichrome composition for the current distribution along the
working electrode with average current density of 12.5mA/cm2.
91
Figure 5.10
Figure 5.14
Figure 6.1
PHI press equipped with a machined vacuum chamber used for
hot embossing of the microfluidic channels
94
Figure 6.2
SEM of molded the polycarbonate using straight channel mold
insert
94
Figure 6.3
SEM of molded the polycarbonate using spiral channel mold
insert
95
Figure 6.4
Cover plate with machined air pockets
97
Figure 6.5
A schematic diagram of the control circuit used to control the
temperature of the PCR device
98
Figure 6.6
A schematic diagram of the experimental setup to test for the
transient response and capture the steady state temperature
contour of the PCR device
Figure 6.7
The comparison between simulated and experimental transient
temperature response
x
100
101
LIST OF SYMBOLS
ρW
Thermal power density, W/in2
ρ
Fluid density
ρR
Specific resistivity
ε
Thermal emissivity
δ ij
Tensor notation
φ
Viscous heat dissipation function
α
α0
αi
Thermal diffusivity
Temperature coefficient of resistivity
Aspect Ratio
∆t
Time difference
ξ
Radiation absorptivity
ν
Kinematic viscosity
ν1
Electrical conductivity
β
1/Temperature
∆T
THot –TCold
σ
Boltzman’s constant
∆P
Pressure difference, psi
αr
Aspect ratio, αr < 1
αtcr
Temperature coefficient of resistance
ρR
Resistivity
∆T
Temperature difference, °C
xi
µ
Dynamic viscosity
a
Channel semi-width, µm
A
Cross-sectional area, m2
b
Channel semi-depth, µm
C
Thermal capacitance
cp , cv
Specific heat capacity constant pressure and constant volume
Dh
Hydraulic diameter, m
e (t)
Effort variable for a bond graph port
f (t)
Flow variable for a bond graph port
g
Acceleration due to gravity, m/s2
hcon
Convective heat transfer coefficient
hr
Radiation heat transfer coefficient
I
Current, A
i
Current density, , A/m2
inv ∆p
Reciprocal of pressure difference, , m2/N
k
Thermal conductivity
Keq
Equivalent thermal conductivity
KP
Proportional gain
KI
Integral gain
KD
Derivative gain
kTh
Thermal conductivity
l
Length of fluid path, m
L
Length of channel , m
Lc
Characteristic length, m
xii
LHW
Length of heater wire, m
m
Mass, kg
Mwt
Molecular weight
n
Number of turns
P
Fluid pressure, N/m2
PE
Electrical power, W
P (t)
Power flowing into or out of a bond graph port
PT
Thermal power, W
PT
Thermal power, W
q
Heat flux
Qcon
Conductive heat transfer, W
Qcv
Convective heat transfer, W
Qrad
Radiation heat transfer, W
Ra
Rayleigh number (Ra),
Rcond
Resistance for conductive heat transfer, Ohm
Rconv
Resistance for convective heat transfer, Ohm
RE
Electrical resistance, Ohm
RE,T
Electrical resistance at temperature T, Ohm
Re
Reynolds number
RT
Thermal resistance, Ohm
Se
Effort source for input into or output from a bond graph
Sf
Flow source for input into a bond graph, m3/s
t
Time, s
T∞
Ambient temperature, °C
xiii
Tref
Reference temperature, °C
Ts
Surface temperature, °C
u
Fluid velocity, m/s
u (x, y)
Fluid velocity in the (x, y) direction, m/s
Umax
Maximum fluid velocity
v
Mean fluid velocity, m/s
Tref
Reference temperature, °C
VE
Electrical voltage, V
xiv
ABSTRACT
The polymerase chain reaction (PCR) is an amplification technique that is used
with deoxyribonucleic acid (DNA) and can produce millions of copies of the starting
material even from single molecules. Since its discovery by Kary Mullis in the mid1980s, PCR has been one of the most important techniques in the arsenal of tools used
for genetic analyses.
For this research a continuous flow polymerase chain reaction (CFPCR) system was
designed, microfabricated from molded polycarbonate, and tested. Finite element
modeling was used to model the system thermal and microfluidic response. X-ray LIGA
was used to fabricate mold inserts for the micro fluidic channels. Polycarbonate was
used instead of PMMA because polycarbonate has a higher glass transition temperature
and is better able to withstand the sustained high operating temperature of the
continuous flow PCR system. Commercial thin film heaters under PID control were
used to supply the necessary heat flux to maintain the steady-state temperatures in the
PCR device.
There are a number of advantages to performing molecular biology on a small
scale. The savings in time and money are the most prominent. A reduction of the cost of
these experiments is accomplished on a variety of levels. First, this continuous flow
microreactor can be mass-produced using existing microfabrication techniques (LIGA).
The small sample sizes require much smaller volumes of reagents. Many of the
enzymes used in molecular biology are quite costly and reducing the volume required
per reaction will significantly decrease the cost of the experiment. By achieving a
quicker and less expensive DNA amplification, DNA analysis could become a
xv
technique used routinely in clinical medicine, leading to a vast improvement in medical
care.
xvi
CHAPTER 1: INTRODUCTION
The polymerase chain reaction (PCR) is widely used as a molecular biology tool to
amplify DNA (deoxyribonucleic acid). DNA is a double-stranded helix of nucleotides,
which carries the genetic information of a cell. PCR is a common method used for
creating copies of specific fragments of DNA by the following three steps, denaturation,
hybridzation (annealing), and enzymatic polymerization (extension) [Stephen, et al.
1994]. PCR can amplify a single DNA molecule into many billions of molecules. The
three steps of the polymerase chain reaction are typically carried out in the same vial, but
at different temperatures.
The first part of the process is to separate the two DNA chains. The target genetic
material must be denatured, i.e., the strands of its helix must be unwound from a doublestrand, into a single-strand by heating the vial to 95 0C. A primer that normally binds to
the DNA cannot operate at such a high temperature, so the vial is cooled to 55 °C. At this
temperature, the primers are "annealed” to the ends of the DNA strands. The final step of
the reaction is to make a copy of the templates. This is known as the extension step where
the DNA synthesizes by extending the primer downs the template DNA, i.e., making
copies of the templates, this is performed at temperature of 72°C.
Since its discovery by Kary Mullis in the mid 1980s [Saiki, at el.1998], PCR has
become one of the most important techniques in the arsenal of tools available for genetic
analyses [Mullis, at el, 1994]. In its short life, PCR has utterly transformed the life
sciences. From the daily practical uses of medical diagnosis to the theoretical framework
of science and from courts of law to field studies of animal behavior. PCR takes analysis
of minute amounts of genetic material to a new level of precision and reliability. It is far
1
simpler and less expensive than previous techniques for duplicating DNA [Saiki, at
el.1998], so PCR has democratized genetic research, putting it within reach of all
biologists, even those with little or no training in molecular biology.
1.1 REQUIREMENTS FOR PCR DEVICES
The polymerase chain reaction involves processing a mixture of buffers, template
DNA, primers (short oligonucleotides that hybridize to specific locations on the target
DNA and define the amplifying region), and a thermostable DNA polymerase. The
temperatures that are typically used in the PCR reaction include 90-94oC for denaturation
of the double-stranded DNA molecule, 50-68oC for hybridization of the primers to the
single-stranded DNA template and 70-75oC for enzymatic polymerization (extension) of
the primers as shown in Figure 1.1. This temperature sequence is then cycled many times
(~10-50) to provide an exponential amplification of the starting material.
According to Innis, (1988) it is possible to have an amplification time per cycle of
approximately 12 s for a 1,000 base amplicon (PCR product). For 25-35 PCR cycles, a
processing time of only 5-7 minutes would be required since the denaturing and
annealing of nucleic acids requires less than 1 s and the extension speed for Taq
polymerase, a thermally stable enzyme used in most PCR reactions, is on the order of 35100 bases per second. A typical PCR device must be capable of obtaining the require
temperature sequence in the smallest possible time. The device must be able to produce
the temperature in the cycle in sequence and maintain them for the required duration in
order to get optimal performance.
2
95 º C
72 º C
55 º C
∆ t
∆ t
4* ∆ t
Figure 1.1: The desired temperature profile for a single PCR cycle
1.2 CONVENTIONAL PCR DEVICE
Conventional PCR is typically performed by thermocycling the PCR mixture in a
thin-walled plastic vial embedded in a temperature controlled aluminum heat sink (also
known as a block thermal cycler). Due to the high thermal capacitances of the material,
the ramp rate of the temperature is typically on the order of 1oC per second for heating
and cooling. For the temperature transition between the 95 0C and 55 0C zones the time
would be approximately 40 sec.
The cycle speed is then limited by the thermal
capacitance of the heat sink and the heat transfer rate between the heat sink and the PCR
sample. To complete this analysis in a conventional PCR block thermal cycler,
approximately 2 hrs is necessary.
1.3 CONTRIBUTION TO PCR DEVICE PRODUCTION
Presently conventional PCR devices are limited by their sizes, rate of heating and
cooling, and cost. Because of theses factors, a study was performed for the manufacturing
of micro-fluidic components capable of performing the same function as the conventional
PCR device. For this research a molded polycarbonate continuous flow polymerase chain
reaction device was design and micro-fabricated. There were several benefits of the
miniaturization of PCR devices.
3
Due to the small sample sizes, much smaller volume of reagents was required.
Enzymes used in molecular biology are quite costly; therefore by reducing the volume
per reaction the cost of experiment was dramatically decrease. The power consumption is
reduced, which also reduces cost. The rate of heating and cooling is much faster because
of its small thermal capacitance and less power to dissipate. Through molding the device
can be produced in large quantities and used as a disposables instrument.
In general the performance of PCR equipment was enhanced, where heating cooling
and material transport were much more efficient, leading to better yield and faster
analysis. By making DNA amplification and analysis quicker and cheaper, the analysis
itself could become more widespread in medicine, leading to an improvement of medical
care. Even though such experimental chip-based devices are not yet ready for public use,
engineers are greatly anticipating the day when doctors can take the devices on the road
and patients will be able to get on the spot diagnosis based on readouts of their DNA.
1 .4 THESIS OUTLINE
The main objective was to design and micro-fabricate a molded polycarbonate
continuous flow polymerase chain reaction device capable of the required temperature
sequence in the lowest possible time. The device must be able to produce the desired
temperature cycles and maintain these temperatures for the time duration required.
This thesis is arranged as follows: Chapter 2-The background of PCR and a review
of the literature on the most resent micro-fabricated PCR devices; Chapter 3-Design and
simulation, where detailed three and two-dimensional numerical modeling of microscale
fluid flows (pressure driven) and heat transfer simulation using ANSYS (v5.7, ANSYS,
Inc., Houston, PA) were performed. The most feasible device configuration was selected
4
based on the simulation results; Chapter 4-Micro-mold inserts were fabricated using the
LIGA microfabrication technique for the hot embossing of polycarbonate with microfluidic channels; Chapter 5-Mathematical models of the thermal characteristics of the
PCR chip were developed and used to determine the thermal power input requirement
and construct the necessary control circuit to maintain the temperatures-required; Chapter
6-The devices were assembled and experiments were performed using a CCD camera to
capture the various temperature contours displayed in a thin layer of liquid crystal applied
over the PCR channel. These images were compared with those obtained from the
simulation. This chapter also includes the discussion, conclusion and recommendations
for future development of this project.
5
CHAPTER 2:BACKGROUND
The polymerase chain reaction (PCR) is a powerful technique, which results in the
rapid production of multiple copies of a target DNA sequence (DNA is a double-stranded
helix of nucleotides). PCR requires a template molecule (the desired DNA) and two
primer molecules to begin the copying process. The primers are short chains of four
different chemical components that make up any strand of genetic material. These four
components commonly known as nucleotides, or bases [Mullis, at el, 1994], are the
building blocks used to construct genetic molecules.
The first step in this process is to separate the two DNA chains of the double helix.
As the two strands separate, DNA polymerase makes a copy using each strand as a
template. The four nucleotide bases, the building blocks of every piece of DNA, are
represented by the letters A, C, G, and T, which stand for their chemical names: adenine,
cytosine, guanine, and thymine. The A on one strand always pairs with the T on the other,
whereas C always pairs with G. The two strands are said to be complementary to each
other [http://www.accessexcellence.org ]
To copy DNA, polymerase requires two other components: a supply of the four
nucleotide bases and something called a primer. DNA polymerases, whether from
humans, bacteria, or viruses, cannot copy a chain of DNA without a short sequence of
nucleotides to "prime" the process, or get it started. So the cell has another enzyme called
a primase that actually makes the first few nucleotides of the copy. This stretch of DNA
is called a primer. Once the primer is made, the polymerase can take over making the rest
of the new chain.Because the products of one PCR cycle can act as templates for the next
6
PCR cycle, the number of new identical molecules produce doubles with each repetition
of the cycle.
The three parts of the polymerase chain reaction are carried out at different temperatures.
The first part of the process separates the two DNA chains in the double helix. This is
done at 95 0C degrees centigrade for about 30 seconds. The primers cannot bind to the
DNA strands at such a high temperature, so the vial is cooled to 55 0C at this temperature;
the primers bind or "anneal" to the ends of the DNA strands for about 20 seconds. The
final step of the reaction is to make a complete copy of the templates. This works best at
around 75 0C. The Taq polymerase begins adding nucleotides to the primer and
eventually makes a complementary copy of the template. If the template contains an A
nucleotide, the enzyme adds on a T nucleotide to the primer. If the template contains a G,
it adds a C to the new chain, and so on to the end of the DNA strand. This completes one
PCR cycle. The three steps in the polymerase chain reaction - the separation of the
strands, annealing the primer to the template, and the synthesis of new strands - take can
be completed in less 12 sec Innis, (1988).
2.1 APPLICATIONS
It is difficult to exaggerate the impact that the polymerase chain reaction (PCR) has
on our world today. The polymerase chain reaction (PCR) is an amplification technique
that is used for nucleic acids and produces billions of copies of the starting material even
from single molecules. PCR is a quick and easy method for generating unlimited copies
of any fragment of deoxyribonucleic acid (DNA). The technique's unmatched ability to
identify and copy the smallest, oldest and even most damaged DNA has proven
7
exceptionally valuable in criminal courts. For instance, PCR is an indispensable tool used
by forensic scientists to conduct a form of DNA-typing commonly known as DNA
fingerprinting [U.S., Congress, Genetic Witness, 1990]. To type DNA, blood is extracted
from a suspect or victim’s clothing. Scientists then study various areas on the DNA
where inconsistencies are found. This helps the scientists to determine if the sample
matches the DNA of a specific person, that of the victim or the assailant. In its earlier
stages, DNA typing was considered controversial. However, laboratory standards have
been established and DNA typing is now accepted as strong evidence in court cases
around the world [U.S., Congress, Genetic Witness, 1990]. Although DNA typing is only
one of many pieces of evidence that can lead to a conviction, it is invaluable in
demonstrating innocence.
2.2 GENETICS
The genetic material of all living organisms including plants, animals, bacterium
and viruses have sequences of nucleotide building blocks that are uniquely present in its
respective species. Correspondingly, complex organisms such as human beings possess
DNA sequences that are uniquely and specifically present only in particular individuals
[Saiki, at el.1998].
These unique variations make it possible to trace genetic material back to its origin,
identifying with precision the species of organism and often a particular member of that
species. A study of such magnitude requires that an ample amount of the DNA under
investigation is available for analysis, which is where PCR comes in.
PCR utilizes natural enzymes known as polymerases. These enzymes are present in
all living things and their job is to duplicate genetic material. Sometimes referred to as
8
"molecular photocopying," it can be used to characterize, analyze, and synthesize any
specific piece of DNA [Stephen, et al. 1994]. It works even on extremely complicated
mixtures by seeking out, identifying, and duplicating a particular bit of genetic material
from blood, hair, or tissue specimens. These tissue samples can be from any species,
even samples that are thousands or possibly even millions-of years old.
2.3 LITERATURE REVIEW OF PCR DEVICE’S
Conventional laboratory techniques for performing PCR are expensive and time
consuming. The performance of micro fluidic devices capable of doing these functions
are significantly better in terms of speed, costs, flexibility and performing multiple
functions on the same device. The proliferation of manufacturing techniques for building
micro-fluidic components has led to a virtual explosion in the development of microscale chemical and biological analysis systems (BioMEMS). Numerous studies have
unveiled chip-sized devices capable of conducting complex analytical procedures
including capillary electrophoresis, and the polymerase chain reaction. Some of the most
publications are tabulated in Table 2.1. This table identifies the different groups, sample
volume used, reaction time, the primary materials used for device construction, and the
number of cycles incorporated on the chip.
To enhance the speed of the PCR reaction, researchers have investigated alternative
methods for heating and cooling by reducing the thermal capacitance of the system. For
example, In 1989 Wittwer and his group used thin-walled capillaries (total volume = 100
µL) containing the PCR mixture, which were then placed in an air-thermal cycler
[Wittwer, et al. 1989 and 1990]. Air was used as the heat source because of its low
density. Air was pumped into the cycling chamber at the appropriate temperature. The
9
air mixed rapidly to achieve a fast temperature transition. Amplifying a 536-bp fragment
from the β-globin gene was accomplished in ~25 mins.
Table 2.1: Cited Works of Prior Micro PCR Devices (see references list for more
details)
Groups
Sample Volume
Reaction time
Material
Cycles
Cheng, et al. 1996
12 µl
30min
Silicon-glass
30
Shoffner, et al. 1996
28 µl
30min
Silicon-glass
35
Woolley, et al.1996
50 µl
26 min 15sec
Si-glass
35
Kalinia, et al. 1997
10 nl
30min
Quartz glass
40
Landers, et al. 1998
5-15 µl
14min
Glass
30
Kopp, et al. 1998
5-20 µl
50min
Glass-copper
30
Friedman, et al 1998
5 µl
20 min
Glass
35
Waters, et al. 1998
12 µl
216 min
Glass
24
Ferrance, et al. 2000
10 µl
7 min 44sec
Silicon
30
Lagally, et al. 2000
10 µl
15 min
Glass
20
Khandurina, et al. 2000
7 µl
20 min
Cr-Glass
10
Giordano, et al. 2000
5 µl
4 min 50 sec
Polyimide
15
Nagai, et al. 2001
1.3 pl -32 µl
120min
Silicon
40
Schneega, et al. 2001
50 µl
35 min
Glass
7
10
Friedman and Meldrum used thin-walled micro-capillaries to contain the PCR mix
and then coated the micro capillaries with a thin-film of indium-tin oxide (ITO) that not
only acted as the heating element, but also a temperature sensor [Friedman and Meldrum,
1998]. They were able to demonstrate amplification of a 777-bp PCR product in 12
minutes (35 cycles, 20.6 s/cycle). Landers and his group also used micro-capillaries as
the reactor, but performed infrared-mediated temperature control, in which a tungsten
lamp generated IR-radiation to excite IR-active vibration modes in the aqueous solvent
[Oda, et al. 1998][ Giordano, 2001]. Amplifying genomic DNA in a sample volume of
5 µl (β-chain of T-cell receptor, 216 base pair amplicon) required approximately 30 min.
Several researchers have focused on improving PCR performance by reducing the
reaction size with a smaller reaction chamber [Cheng, et al. 1996]. For example, some
scientists are using capillaries with hot-air thermal cyclers [Woolley, 1996] or infraredmediated heating [Huhmer and Landers, 2000] as their heat source. Scientists are also
directly integrating PCR reactors into microchips. Several groups have recently been
successful at including on a single chip both the PCR reactor and the capillary
electrophoresis device for subsequent analysis of the reaction products [Burns, et al. 1996
&1998],
A flow through type of PCR system based on microfabrication technology was a
contribution of Kopp and Manz (1998). A glass chip with a 40 µm by 90 µm by 2.2 m
channel was built using a lithographic technique. The PCR mixture was pumped by a
syringe along the channel through three different temperature zones at flow rates ranging
from 5.8 to 72.9 nL/s. The PCR products were collected and analyzed on a
polyacrylamide gel stained with ethidium bromide. The continuous flow PCR chip
11
showed less cross-contamination than a stationary-tube format because of continuous
washing. The flow through chip could be reused without significant cleaning. Their
results indicate that the PCR product could be visualized at a total processing time of 50
minutes producing a cycling time of 7.5 s/cycle.
Most recently, Schneegass created a silicon and glass chip, flow through PCR
device [Schneegass, et al. 2001]. The silicon and glass chip with a 250 µm by 100 µm by
1.512 m channel etched in glass. A syringe pump was used to drive the PCR mixture
through the chip. For each experiment a sample volume of 10 µl was used. The optimal
flow rate was 0.0167 µl/s, which give a total time of 35min for 20 cycles. Their
amplification was 606 bp, which was larger than Kopp’s 176 bp.
12
CHAPTER 3: DESIGN OF A TEST DEVICE AND FINITE FLEMENT (THERMAL)
ANALYSIS
3.1 INTRODUCTION TO THE CONTINUOUS FLOW STRAIGHT CHANNEL PCR
DEVICE
The continuous flow straight channel PCR device was based on the movement of
small packets of fluid between constant temperature zones in the manner of Kopp (1998).
This significantly simplified the temperature control problem, since the transient heat
transfer in the device was confined to start-up and the effect of the thermal capacitance of
the components on the response time as a packet moved through the PCR cycle was
eliminated.
A steady state thermal analysis was performed to evaluate the temperature profile
along the PCR channel. The principal underlying assumptions were that the heating
elements could be configured to create distinct nearly-constant temperature zones of the
correct length in the fluid channel and that the fluid packets would attain equilibrium
within a short distance, relative to the length of the section, of entering each constant
temperature zone. A series of finite element simulations using ANSYS/CFD-FLOTRAN
(vers. 5.7, ANSYS, Inc., Canonsburg, PA) was carried out to evaluate these assumptions
and to define a prototype device configuration for experimental evaluation.
Different materials were investigated for their workable temperature range and
formability. Also, different methods of thermal power input were analyzed in order to
determine method of heating, which would lead to the simplest fabrication process and
reduce production cost. The effects of increasing fluid velocity were investigated since
high speed processing is a goal for clinically useful devices.
13
3.2 MATERIAL SELECTION
Polycarbonate was selected as the material for the PCR module. The desired
maximum steady-state temperature was 95 oC in the denaturation zone. This temperature
was too high for the PMMA (glass transition temperature approximately 100 oC
[http://www.matweb.com, 2002] which was previously used to mold capillary
electrophoresis devices by our group [Soper et al., 2000]. For polycarbonate the glass
transition temperature is typically in the range of 145 oC – 148 oC, which was suitable for
the higher operating temperature, while the thermal conductivity of the two materials was
essentially the same (PMMA 0.19 W/m–oK, PC 0.22 W/m–oK) [1]. The low thermal
conductivity
compared
to
glass
(1.1 W/m–oK
and
silicon
(124 W/m–oK
)
[http://www.matweb.com, 2002] meant that less power would be required to maintain the
steady-state temperatures needed in the module. Initial studies also showed that
microfluidic channels could be hot embossed in polycarbonate and sealed by thermal
bonding of a polycarbonate cover plate.
3.3 DEVICE CONFIGURATION
The test device consisted of a single straight 50 µm by 150 µm fluidic channel. This
continuous flow polymerase chain reaction (CFPCR) device was designed, and fabricated
from molded polycarbonate. The device had two cover plates, which incorporated air
pockets, electrical resistive heaters, and a hot embossed microfluidic channel. Figure 3.1
shows the cross-section of the test device configuration. All of the dimensions of the test
device configuration can found in Chapter 6 and Appendix C.
The continuous flow PCR was based on the movement of small packets of fluid
analyte between constant temperature zones. These isothermal zones, through which the
14
fluid flows, are very important for achieving and maintaining the desired profile. There
were two aspect of maintaining constant temperatures in the isothermal zones. The first
task was to minimize heat transfer to the environment. In the future, the PCR device will
be stacked with other modules in forming multi functional instruments; therefore it is
imperative that the heat transfer to the environment is minimized. The second task was to
determine the optimal location for the resistive heaters used as the thermal energy source
in each zone. Spacing between heaters affects the width of the transition band between
temperature zones.
Figure. 3.1: A cross-sectional schematic view of the straight channel test device (not to
scale). Configuration is similar to the spiral device
15
3.4 FINITE ELEMENT ANALYSIS (ANSYS /FLOTRAN)
Finite Element Analysis (FEA) is a way to simulate loading conditions on a design
and determine the response to those conditions. The model is composed of discrete
building blocks called elements. Each element has exact differental equations that
describe how it responds to a certain load. The sum of the responses of all of the elements
in a model gives the total response of a system.
In order to obtain a correct solution using the finite element analysis package,
(ANSYS vs. 5.7, Canonsburg, PA), the following were completed: (1). The equation (or
system of equations); boundary and initial conditions were clearly defined to ensure that
a well-defined problem was formulated; (2). An element type was chosen which defined
the approximation function used in the solution, (3). A mesh was created that adequately
defines the regions where large changes in the solution were expected, which allows the
boundary conditions to be properly imposed.
The FEA package (ANSYS) was capable of formulating the finite element
algorithm that was used to solve the resulting system of algebraic equations. It also
calculates the error in the approximation to determine if the solution was converging or if
a more refined solution was needed (meaning a reduction in mesh size).
Finite Element Analysis (FEA) was carried out separately on three different
sections of the device. First it was used to analyze the air pocket for both heat convection
and thermal radiation effects. Second the method was used to do two dimensional heat
conduction analyses for preliminary studies of heater placement. Third coupled thermal
16
and fluid analysis of the test device was carried out. This analysis was the combination of
steady-state heat conduction through the solid region of the model and forced convection
through the micro channel.
3.5 STEADY-STATE THERMAL
A steady-state thermal analysis calculates the effects of steady thermal loads on a
system or component. Steady-state thermal analysis can be used to determine temperature
distribution. Such loads include, convection, radiation, heat fluxes, heat generation rates,
and constant temperature boundaries. Steady-state thermal analysis may be either linear,
with constant material properties, or nonlinear due to radiation. If the material properties
of the system are temperature dependent then the system may be nonlinear. Radiation
effects are nonlinear because it varies with the fourth power of the absolute temperature.
3.6 ANALYSIS OF THE AIR POCKETS
Air pockets were incorporated in the design to maximize the thermal insulation
above and below the heated microfluidic channels. Air is a poor conductor of heat and the
convective effects could be neglected.
It is possible to obtain heat conduction and thermal radiation across the air pocket
by neglecting the effects of heat convection. For a sub-layer of fluid of thickness L
confined between two parallel, isothermal plates and heated from below, the heat transfer
through the layer is governed by the difference in temperature between the two plates,
(TH
− TC ) . For smaller values of (TH − TC ) , the fluid is stationary and the heat transfer
through the layer is assumed to be by conduction only. However, if (TH − TC ) is
increased to a critical value, the fluids will become unstable, and a cellular flow pattern is
established. The circulation of fluid within a cell convects warm fluid upward and cold
17
fluid downward. To accommodate the higher rate of heat transfer across the layer, the
isotherms adjacent to each plate move closer together. If (TH − TC ) is further increased,
there are transitions to increasingly more complex flow patterns until a chaotic turbulent
motion replaces cellular flow. The core of the fluidic layer is then almost isothermal, with
the major temperature variations confined to a very thin viscous sub-layer adjacent to
each plate, where viscosity dampens the turbulent motions [Mills 1999].
TC
TC
TH
(a). Stationary fluid with evenly spaced
isotherms because of conduction only.
TH
(b). Cellular flow with its streamlines
TC
TC
TH
TH
(c). Isotherms
(d). Turbulent mixing isotherms close
together in the viscous sublayer
Figure 3.2: Natural-convection flow regimes for a layer of fluid between tow horizontal
isothermal plates. The temperature different (TH − TC ) increases from regime
(a) through to regime (d) [Mills 1999].
The criterion for instability and the onset of cellular convection is a critical value of
the Rayleigh number (Ra), which is a dimension less parameter
18
Ra = gβ∆TL3 ρ 2 C p / kµ = 1708
3.1
where:
g = gravitational acceleration (m/s2),
β=1/T (1/K), ∆T=THot –TCold (0C)
L= thickness of the cavity (m),
ρ = density (kg/m3),
µ = viscosity (kg/m s),
k = thermal conductivity (W/m K)
and Cp = specific heat (J/kg/K)
Finite element simulations using ANSYS (CFD-FLOTRAN) (vers. 5.7, ANSYS,
Inc., Canonsburg, PA) were carried out to evaluate the assumption that heat transfer
across the insulating air layer was primarily by conduction and radiation.
3.7 PROBLEM DESCRIPTION
The simulation evaluated a buoyancy-driven flow in a square cavity (air pocket)
heated from below. The thermal conductivity of the air and the movement of the fluid
governed the heat transfer through the layer. A laminar, steady-state analysis that used
the FLUID141 element was performed. FLUID141 is two-dimensional element used in
FLOTRAN to model transient or steady-state fluid/thermal systems that involve fluid
and/or non-fluid regions [ANSYS/FLOTRAN 5.4, Analysis Guide 1997].
19
T reff h ∞
Polycarbonate
Air W
g
L
L
Q
Figure 3.3: The model of the rectangular cavity simulated for the
air pocket
The dimensions of the simulated cavity were 20mm wide x 1mm deep and the
operating conditions were reference temperature of 300K,a reference pressure 1.0135 х
103 Pa, and gravitational acceleration 9.81 m/s2. The units for all air properties are in the
SI system. No slip boundary conditions were assumed for the walls (Vx = Vy = 0). At the
two ends, adiabatic (meaning
∂T
= 0 ) boundary conditions were assumed because the
∂x
ratio of the length to the depth of the cavity was 20:1 and the end effects could be
neglected. A heat source Q = 400W/m2K was used to ensure that the air was in the cavity
was heated to our maximum working temperature of 950C, which was also enough to
induce a cellular flow pattern if possible. The top surface of the polycarbonate was
exposed to natural convection where h∞, the convective heat transfer coefficient, was
taken as 50W/m2, the value normally used for conservative design.
20
Figure 3.4
shows a stationary fluid with evenly spaced isotherms because
convection is not a factor. This is in agreement with the case shown in the schematic in
Figure 3.2 (a) and validates the assumption of conduction dominated heat transfer.
Figure 3.4: The contour of the temperature distribution across the heated cavity
Figure 3.5: The contour of the pressure increase within the heated cavity
The pressure increase on the inside of the pocket was another concern. This may
cause expansion, which would induce stress on the bonded joints and the heating
elements. Based on the results obtained from the simulation, shown in Figure 3.5, the
maximum pressure change would be negligible when compared to the reference pressure
of 1.0135 х 103 Pa.
21
3.8 THERMAL RADIATION AND CONDUCTION
The dominant heat transfer mechanisms through the air pocket were radiation and
conduction. The heat flow due to radiation varies with the fourth power of the absolute
temperature of a body. In heat conduction, which is governed by Fourier’s Law, the local
heat flux is proportional to the negative of the local temperature gradient, which is linear.
In order to combine these two parallel but separate heat flows, the radiation heat
transfer was linearized from:
(T14 −T24 )
Q12 = Aσ
1 1
+ −1
32
ε1 ε 2
where ε is emissivity of each wall, σ is Stefan-Boltzmann constant (5.67х10-8 W/m2
K4) , A is the surface area and T1, T2 are the temperature of the top and bottom surface
respectively. The emissivity of polycarbonate is ε = 0.9 ≈ 1 therefore Equation 3.6 is
simplified to:
Q12 = Aσ (T14 −T24 )
3.3
According to Mills [1999] if the different between T1 and T2 is small when
compared to the absolute temperature, Equation 3.3 can be linearized to yield an
approximate radiation heat transfer coefficient of
hr = σ 4Tm3
3.4
where Tm is the mean temperature between T1 and T2 and heat flow between each
radiating surface is now:
Q12 = Ahr (T14 −T24 )
3.5
22
By using the electrical analogy of heat transfer for two parallel resistors, the total
heat resistance across the air pocket was formulated as shown in Figure 3.6 where:
R cond =
LT
k Th A
R rad =
1
hrA
3.6
3.7
and the total heat resistance across the air pocket is
R total =
LT
A(k Th + h r L T )
3.8
where LT is the characteristic length between the two radiating surfaces. This resistance
was used to calculate the equivalent thermal conductivity for the air pocket, which is
Keq, = (kTH + hrLT). For air at a temperature of 950C this equivalent thermal conductivity
is 0.07 W/m2K, which is one third of the thermal conductivity of polycarbonate of
(0.23W/m2K), which makes air the better insulator.
Rrad
q
Rcond
LT
Figure 3.6: These are two parallel resistors, radiation and conduction across the air pocket
23
In order to validate the assumptions used in the above radiation analysis, laminar,
thermal, steady state FLOTRAN analysis was performed using the model shown in
Figure 3.3. The radiosity solution method was used with surface-to-surface radiation for
the buoyancy-driven flow in the heated cavity. All the walls of the cavity were taken to
be radiating surfaces because the cavity was fully closed. The walls of the cavity were of
polycarbonate which an emissivity of 0.9. The final temperature distribution in the cavity
was affected by the presence of the radiating walls in the fluid domain and the heat
conducted through the fluid.
A steady-state ANSYS thermal conduction simulation was done for the same air
filled cavity. Treating the air as solid with an equivalent thermal conductivity of Keq =
(kTH + hrLT). The heat load and the boundary conditions remained the same as in the
convective analysis. The temperature distributions across the cavity for both simulations
were compared to each other in order to observe the differences (see Figure 3.7). Figure
3.7 shows that the difference between the two model predictions was s less than 1% at
any point across the cavity, therefore the assumptions and linearized model were
considered valid.
3.9 FINITE ELEMENT ANALYSIS FOR PRELIMINARY SIMULATIONS
All of the preliminary simulations were carried out for conduction heat transfer
only, meaning that forced convection through the channel was neglected because of slow
movung fluid (2mm/s) and small fluid volume which convects negligible thermal energy.
In doing so, the problem was simplified significantly. The problem was further simplified
24
Temperature Across the Air Pocket
370
365
Temp of Cond only with equivalent (Ke)
Temp of Radiation and Conduction
Temperature(K)
360
355
350
345
340
335
0
0.2
0.4
0.6
0.8
1
1.2
Channel Depth(m)
1.4
1.6
1.8
2
-3
x 10
Figure 3.7: The comparison of the temperature distribution across the air pocket, when
simulated with radiation and conduction and conduction only with the
equavalent thermal conductivity. The enlarged section shows clearly the
temperature differences.
by setting the temperature on the outer surface of the device to a constant 22 0C.
ANSYS was used to simulate two-dimensional steady-state conduction heat transfer
across the section (see Figure 3.8).
25
∂T
∂  ∂T  ∂  ∂T  ∂  ∂T  •
 +  k
 + q = ρc
k
 +  k
∂t
∂x  ∂x  ∂y  ∂y  ∂x  ∂z 
3.9
Equation 3.9 is the general heat conduction equation with variable thermal
conduction in the rectangular coordinate system [Kakac, 1993]. Since there was no heat
•
generation and steady state condition: q = 0 and
∂T
= 0 respectively. The equation
∂t
simplified to:
∂ 2T ∂ 2T ∂ 2T
+
+
=0
∂x 2
∂y 2
∂z 2
3.10
∂ T ∂ T
+
= 0 → for − 2 D
∂x 2
∂y 2
2
2
The selected ANSYS element for the steady-state heat conduction model was
PLANE55. It can be used as an axisymmetric element with a two-dimensional thermal
conduction capability. The element has four nodes with a single degree of freedom,
temperature, at each node. The element is applicable to a two-dimensional, steady-state,
or transient thermal analysis.
3.10 H EATER POSITIONING AND LOCATION
Simulations were used to determine the optimal number and locations of the
heaters needed to obtain the constant temperature zones for the PCR cycle as shown in
Figure 1.1. The three functional regions represent nominally constant temperature zones
in the continuous flow device. There were three temperature transitions: (1) from 950C to
550C; (2) from 550C to 720C; and (3) from 720C to 950C. Two principal questions were
addressed. First was to determine the spacing between the heaters that would maintain the
26
temperature in each zone with the narrowest transition band. Second was an evaluation of
whether heaters would be needed on the top only or on both the top and bottom of the
device, in order to ensure a uniform temperature profile across the microchannels.
The critical section of the device was the transition from 950C to 550C therefore
preliminary simulations was carried for this section only. A straight section of the device
containing two constant temperature zones and one transition zone was used to the model
the system. Two-dimensional, steady-state heat transfer was assumed. The models were
executed in ANSYS (vers 5.7). The heaters were represented by constant heat flux as the
sources. Each cycle for PCR device was repeated at 95 0C, 55 0C and 72 0C, therefore the
inlet and outlet temperature of the critical section (95 0C to 55 0C) was considered to be
72 0C.
T=220C
T=720C
T=72 0C
T= 220C
Figure 3.8: The cross-section of simulated straight section of the device,
which contained two constant temperature zones and one transition
zone
In order to obtain a short PCR cycle time, the durations of the temperature
transitions between the zones must be minimized. The spacing between the heaters,
influences the temperature gradients and the transition time between each temperature
zone. Due to the goal of a small sized device, the spacing between heaters must also be
27
kept to a minimum provided that it does not cause any excessive temperature nonuniformity in the transition region.
Heat conduction away from the channel in the unheated section between
temperature zones was the only source of cooling between the denaturation (95 0C) and
hybridization (550C) sections. There was a concern that, in relying on only ambient
cooling, the distance between heaters required for the temperature to decrease to the
desired level would be unacceptably long. Therefore, the use of metal cooling fins in the
transition zone to accelerate the temperature drop was evaluated. ANSYS was used to
simulate two modes, one with fins and the other without fins. The simulated results are
shown in; Figure 3.9 (a) for the channel with cooling fins, and in Figure 3.9 (b) for a
system without fins.
With the fins, the temperature in the transition zone dropped below the
hybridization temperature before recovering. When the fins were removed, a monotonic
decrease in temperature from the denaturization (95 0C) level to the hybridization (550C)
was observed. Therefore the fins were eliminated from the design, which simplified the
fabrication and assembly of the final device. By iteratively varying the positions of the
heaters, an acceptable temperature profile across the transition zone was obtained.
The objective of this device was to have as close to a constant temperature in each
zone as possible and a uniform temperature profile across the microchannel so that all of
the analyte would be exposed to the same temperature-time profile. Models of two
different heater configurations were simulated, with heaters on one side or on both sides
of the channel. The simulation results for the two different models are shown in Figure
3.10 (a) and (b). The device with heaters on both sides showed a more uniform
28
temperature distribution across the channel, but the final assessment of each
configuration will be made after the simulation of the final design.
(a)
(b) contour obtained for the device, each color
Figure 3.9: ANSYS simulated temperature
represent approximately a 100C temperature bond where different colors
showing the temperature variation. (a) The device with fins and (b) The
device without fins
.
(a)
(b)
Figure 3.10: ANSYS simulated temperature contour obtained for the device, where
different colors showing the temperature variation with a 100C temperature
bond. (a) The device with heaters one top and bottom and (b) The device with
heaters on top only.
29
For the simulation of the final device a combination of steady-state heat conduction
and forced convection through the microchannel was used. Cooling fins were eliminated
from the design. However both heater configurations, with heaters on one side and
heaters on both sides of the channel were simulated. The model with heaters on one side
would be easier to fabricate, particularly for modular components, so the choice would be
made after the final simulations were completed.
3.11 FLUID ANALYSIS OF THE TEST DEVICE
A fluid analysis was preformed to investigate the entrance effect of the velocity.
This was distance the fluid traveled along the channel before it would be
hydrodynamically fully developed. If the entrance effect could be neglected, the flow
would be simulated as being fully developed flow.
To validate the simulation, a comparison was made between the analytical and
simulated velocity profiles for three different mesh sizes. The mesh that gave the closest
approximation to the analytical solution was used for the final device simulation.
For a mean velocity of 2 mm/s, the cycle duration was 30 sec of which the
respective time for denaturation (95oC), hybridization of the primers (55oC) and
enzymatic polymerization (extension72oC) are ∆t, ∆t and 4∆t respectively, hence ∆t was
5 sec.
Water has characteristics similar to the analyte so the properties of water were
assumed to be that of the actual mixture. The average steady-state temperature of the
working fluid is 740C therefore all of the properties were taken at this temperature.
With the above parameters and fluid properties, the Reynolds (Re) Number and the
30
Table 3.1: The design parameters and specifications
Zones
Time
Dimensions
Denaturation (95oC)
5 sec
Length (6 m)
Hybridization (55oC)
5 sec
Width (50µm)
Extension (72oC)
20 sec
Depth (150µm)
Total time
30 sec
Volume = 0.45µl/cycle
Material properties
Symbols
Values
Kinematic viscosity:
ν
0.44*10-6m2/s
Density
ρ
975kg/m3
Thermal diffusivity=
α
1.6*10-7m2/s
hydraulic diameter were calculated. For a rectangular duct, the equivalent hydraulic
diameter is Dh =
u * Dh
4wd
, and Reynolds (Re) Number Re =
. The calculated
2 * (w + d )
ν
values of hydraulic diameter, and Reynolds’s (Re) Number were 75µm and 0.351,
respectively. This showed that the flow was laminar because Re < 2300. For laminar
flow, the hydrodynamic development length in entrance region is:
0≤
Le
≤ 0.05
Dh Re
3.11
In this application the velocity development length was calculated to be 1.32 µm, which
was negligible when compared to the total length of the channel. Fully developed flow
was assumed for the simulation.
31
y
2b
z
2a
Figure 3.11: Cross section of the rectangular micro duct
Three-dimensional analyses for fluid flow and heat transfer through the rectangular
duct were carried out for fully developed laminar flow. Consider the cross section of a
rectangular duct shown in Figure 3.11 with the flow direction along the x axis
perpendicular to the plane of the paper.
According to Dryden (1959), such a section is characterized by its aspect ratio, b/a.
For the rectangular section where –a ≤ y ≤ a and –b ≤ z ≤ b, the velocity profile is given
by:
 cosh(iπz / 2a)  cos((iπy / 2a )
16a 2  dp  ∞
−  ∑ (−1) ( i −1) / 2 1 −
u ( y, z ) =
×
3 
µπ  dx i =1,3,5,.
i3
 cosh(iπb / 2a) 
3.12
where dp/dx is the pressure gradient given in terms of the mean velocity, which is:
 dp 
 dx 
−1
=
a2
3

1   192a ∞
(iπb / 2a) 
 −
 1 − 5 ∑ tanh

i5
 u men   π b i =1,3,5,.

3.13
Matlab 2000 (The Mathworks, Natick MA) was used to solve the above equations
and the result is shown in Figure 3.12.
32
x 10
Velocity Distribution In the PCR Channel
-3
4
3.5
3
Velocity
2.5
2
1.5
1
0.5
0
-2
x 10
-5
0
W idth of Channel
2
4
6
2
0
-2
Depth of Channel
-4
-6
x 10
-5
Figure 3.12: Three-dimensional velocity profile of the pressure driven flow in the PCR
channel
The flow rate is given by:
Q=
4ba 3  dp   192a ∞
(iπb / 2a ) 
 −  1 − 5 ∑ tanh

3  dx   π b i =1,3,5,.
i5

3.13
The maximum velocity (Umax) at the center of the section was 3.727 mm/s and the
corresponding pressure gradient (dp/dx) at the average velocity of Umean 2 mm/s was
-1.2152*104 (Pa/m).
33
3.12 COUPLED ANALYSIS OF THE TEST DEVICE
Thermal Analysis Element type: The coupled analysis of the test device was the
combination of steady-state heat conduction through the solid region of the model and
forced convection through the microchannel. Equation 3.9 was solved in threedimensions using ANSYS and the resulting temperature distribution was used as the
boundary condition for the fluid simulation. The element selected was SOLID187. This
element is a higher order 3-D thermal element. It has a quadratic displacement behavior
and is well suited to modeling irregular meshed objects. Ten nodes defined the element,
which had three degrees of freedom at each node: translations in the nodal x, y, and z
directions. Temperatures may be input as element body loads at the nodes. If all of the
corner node temperatures are specified, each midside node temperature will default to the
average temperature of its adjacent corner nodes.
3.13 FLOTRAN (CFD) ELEMENT TYPE
Flotran CFD uses the finite element method for fluid flow analysis.
Computational fluid dynamics uses computers to solve the equations that described the
motion of fluids, i.e., both liquids and gases. The equations that govern the motion of a
fluid are based on the principles of conservation of mass, momentum and energy.
The most general form of the mass conservation or continuity equation, using
indicial form, is:
∂ρ ∂ρu j
+
=0
∂t
∂x j
3.14
34
where ρ is the fluid density , uj is the velocity (here j=1.2.3 denoted u, v and w in the x, y
and z direction, respectively), and t denoted time. [Kundu, 1990] If the fluid is
incompressible
Equation 3.9 reduces to
∂u j
∂x j
=0
The conservation of momentum equation for an incompressible, Newtonian,
viscous fluid can be written as:
 ∂u i
ρ 
 ∂t
+uj
∂u i
∂x j

 = − ∂p + ∂

∂xi ∂x j

 ∂u i
µ
 ∂x
j


 + ρBi


3.15
where p is the static pressure, µ is the molecular viscosity of the fluid and ρBi are the
gravitation and body forces in the i direction. Kundu [1990]
The energy conservation equation is based on the first law of thermodynamics.
Conservation of energy for an incompressible fluid is written in terms of internal energy
per unit mass, as:
 ∂T
ρcv 
 ∂t
+uj
∂T
∂x j


 = ∂  k ∂T
 ∂x  ∂x
j 
j


 +φ + f


3.16
where φ is the viscous heat dissipation function and is defined as
  ∂u
∂u j 
∂u
 ∂u
 + λ k δ ij  i
φ =  µ  i +
∂x k  ∂x j
  ∂x j ∂xi 
3.17
where k is the fluid thermal conductivity and f if a volumetric heat source or sink. In
many applications involving incompressible fluids, the dissipation term is very small and
is often neglected. [Dryden, 1956]
The element selected was FLUID142. This is a 3-D element that can be used to
model transient or steady-state fluid/thermal systems that involve fluid and/or non-fluid
35
regions. The conservation equations for viscous fluid flow and energy are solved in the
fluid region, while only the energy equation is solved in the non-fluid region. The
FLOTRAN CFD element can be used to solve for flow and temperature distributions
within a region.
For the FLOTRAN CFD elements, the velocities are obtained from the
conservation of momentum principle, and the pressure is obtained from conservation of
mass. The temperature is obtained from the conservation of energy. FLOTRAN used a
segregated sequential solver algorithm, the matrix system, derived from the finite element
discretization of the governing equation each degree of freedom is solved separately. The
flow problem was nonlinear and the governing equations were coupled. The sequential
solution of all the governing equations, combined with the update of any temperature or
pressure dependent properties, constitute a global iteration ANSYS {vs. 5.4 Thermal
Analysis Guide, 1197].
3.14 PROBLEM CONFIGURATION
The model system was comprised of two cover plates, which were exposed to
convective heat transfer. The convective heat transfer coefficient was assumed to be
50w/m2k and the ambient air temperature was 220C. There were six air pockets, two for
each temperature zone, one at top and one at the bottom. The core of the device consisted
of a microfluidic channel sandwiched between polycarbonate and the air pockets.
Constant heat fluxes were used to simulate the electrical heaters as the heat source. Water
was used as the working fluid.
Figure 3.13 is a cross-sectional schematic view of the model with the applied
boundary conditions. The ratio of the width of the channel to the width of the device was
36
on the order of 1000:1, so the heating surface was assumed to be infinitely long. This led
to the assumption that the boundaries in the z direction were adiabatic. By modeling the
device as a closed loop, which was cut through the 720C zone and straightened out, the
boundary condition at each end was also understood to be adiabatic (in the x-direction).
∂T
=0
∂z
∂T
=0
∂z
(a)
(b)
Figure 3.13: Two cross-sectional views of the model: (a) In the xy plane, and (a) in the yz
plane.
37
Figure 3.14: This is a three-dimensional view of the simulated section of which two
views are shown in Figure 3.13
3.15 MESH SIZES
Three different mesh sizes were evaluated to ensure that the simulated results
obtained are independent of element size. These different meshes were drawn using
ANSYS, with different distribution of cells along the channel wall and the temperature
transition zones. The model was meshed with finer cells close to the channel wall and
between each transition zone, where the respective velocity gradient and thermal
transition length are important.
Mesh (1).
Mesh (2).
Mesh (3).
Figure 3.15: The cross sections of the right end for each of the three different mesh sizes
of the fluid channel and its surrounding solid region.
38
Transition section
Mesh (1)
Mesh (2)
Mesh (3)
Figure 3.16: The cross section of the three different mesh sizes for each temperature
transition zones
Table 3.2: The different element dimensions of the three mesh configurations
Mesh
Micro-channel (µm)
Air Pocket (µm)
Transition (µm)
1
(10 х 5 х 20)
(10 х 10 х 20)
(10 х 10 х 20)
2
(40 х 40 х 50)
(40 х 40 х 50)
(40 х 40 х 50)
3
(10 х 10 х 10)
(20 х 20 х 20)
(20 х20 х 40)
39
3.16 FINITE ELEMENT SIMULATION RESULTS
In order to verify the validity of the numerical simulation, a comparison was done
between the analytical and the numerically simulated velocity profiles for the three
different mesh sizes. The closest velocity profile to the analytical solution was Mesh (1),
with a maximum velocity difference of 0.005mm/s, which was negligible.
After the validation of the numerical simulation with the analytical results for the
velocity profile shown in Figure 3.17, ANSYS was used to evaluate the thermal transition
length between temperature zones using Mesh (1).
4
x 10
-3
Comparison of numerical and analytical velocity profiles
3.5
3
Velocity m/s
2.5
Analytical velocity profile
Mech (1)
Mech (2)
Mech (3)
2
1.5
1
0.5
0
0
1
2
3
4
5
6
Channel depth from centerline (m)
7
8
x 10
Figure 3.17 (a): The comparison of the analytical velocity profile to the
simulated result obtained from ANSYS
40
-5
-3
x 10
Comparison of numerical and analytical velocity profiles
3.74
Velocity m/s
3.72
3.7
3.68
Analytical velocity profile
Mech (1)
Mech (2)
Mech (3)
3.66
3.64
7.5
8
8.5
9
Channel depth from centerline (m)
9.5
-6
x 10
Figure 3.17 (b): An enlargement of the comparison of the analytical and
simulated velocity profiles.
The first two simulations were done with zero fluid velocity for two different heater
configurations. Single-sided heating as shown in Figure 3.18 gave an adequate
temperature profile with longer temperature transition (see Table 3.3). The extent to
which the amount of temperature non-uniformity that can be allowed is still unknown and
experiments will be needed to determine this.
41
Table 3.3: The temperature transition length between each zone.
Temperature transition zones
72-950C
95-550C
55-720C
One side heating temperature transition
length
1.8 cm
1.5 cm
1.8 cm
1.1 cm
0.8 cm
Two side heating temperature transition 1 cm
length
Temperature vs Channel Length
370
365
360
Heaters on one side
Heater on both sides
Temperature(K)
355
350
345
340
335
330
325
0
0.01
0.02
0.03
0.04
Channel Length (m)
0.05
0.06
Figure 3.18: The comparison of the temperature distribution along the channel for the
model with heaters on one side and with heaters on both sides.
42
(a)
(b)
Figure3.19: (a) The temperature contour across the device with heaters on one side (b)
The temperature contour across the device with heaters on both sides
The objective was to achieve a simulation as close as possible to the real device, by
applying temperature constraints at specific locations, where the thermocouples for the
control system were located. These temperatures were kept at 950C, 550C and 720C,
respectively, by varying the different heat fluxes (q95 q55 and q72) to reflect the closed
loop control of the temperature at those locations. Table 3.4 and Figure 3.20 shows the
variation of the heat fluxes and the temperature distribution for the different velocities.
43
Table 3.4: The variation in the input heat fluxes for the different velocities to obtain
the applied temperature constraint.
Heat flux (W/m2)
q95 (W/m2)
q55 (W/m2)
q72 (W/m2)
1900
1925
1947
1970
2000
2100
100
85
70
48
15
0
900
920
927
935
942
950
Velocities (mm/s)
0
1
2
3
4
5
Tem perature vs Channel Length for different Fluid V eloc ities
370
365
V=0
V=1
V=2
V=3
V=4
V=5
360
Temperature(K)
355
350
m m /s
m m /s
m m /s
m m /s
m m /s
m m /s
Velocity
increase
345
340
335
330
325
0
0.01
0.02
0.03
0.04
Channel Length (m )
0.05
0.06
Figure 3.20: The variation of temperature distribution along the microfluidic channels,
showing how increasing velocity affect the temperature profile, with the
design temperature constraint.
44
One of our design objectives was to increase the speed of the amplification. Higher
average velocity means higher throughput faster results. The study of increasing velocity
through the channel was performed to determine what the costs were for this particular
device configuration.
The residence time in each zone was defined as the time spent by the fluid in each
isothermal zone. With a temperature bandwidth of ± 50C, the same used by Kopp [1998]
for the 720C and the 550C zones and 0 to -40C for the 950C, the residence times in each
zone were calculated and the ratios of the actual to the design time spent in each zone
were evaluated:
∆t 72 ∆t 95
∆t 55
,
and
The residence time in each of the temperature
3∆t ∆t
∆t
transition zones was quantified by the ratios of the transition time to the design time (∆t =
5sec) for a mean velocity of 2 mm/s, which were:
∆t 72−95 ∆t 95−55
∆t 55−72
,
and
. These
∆t
∆t
∆t
times are shown on Figure 3.21 where: ∆t72 , ∆t95 and ∆t55 , are the time the fluid spent in
the 720C, 950C and 550C zones, respectively. Also ∆t72-95 , ∆t95-55 and ∆t55-72 , are the time
the fluid sent in transition from 720C to 950C, 950C to 550C and 550C to 720C . The
estimated results from the simulation were plotted versus the respective velocities and are
shown in Figure 3.22 and Figure 3.23. Figure 3.22 shows time spent in each isothermal
zone as a function of velocity and Figure 3.23 also shows the time spent in temperature as
a function of velocity. After analyzing the velocity effect on of temperature profile along
the channel, the assumed maximum velocity was determined. Plots of the temperature at
the channel wall and center are shown in Figure 3.24.
45
Temperature vs time for V=0 mm/s
370
-50C
365 ∆t72-950C
360
Temperature (K)
355
∆t950C
350
345
∆t55-720C
340
∆t720C
0
±5 C
∆t550C
335
∆t720C
∆t95-550C
330
325
± 50C
0
5
10
15
Time (sec)
20
25
30
Figure 3:21 Residence times ∆t72 , ∆t95 and ∆t55 , are the time the fluid spent in the
720C, 950C and 550C zones, respectively, while ∆t72-95 , ∆t95-55 and ∆t550
0
0
72 , are the times the fluid spent in transition from 72 C to 95 C, 95 C
to 550C and 550C to 720C.
46
1
0.9
Actual time/Design time
0.8
0.7
0.6
dt(72C)/3dth
dt(95C)/dth
dt(55C)/dth
0.5
0.4
0.3
0.2
0.1
0
1
1.5
2
2.5
3
3.5
Velocity (mm/s))
4
4.5
5
Figure 3.22: The ratio of the residence times of the fluid in the 720C, 950C and 550C
zones as function of mean velocity.
1.6
Actual time/Design time
1.4
dt(72-95C)/dth
dt(95-55C)/dth
dt(55-72C)/dth
1.2
1
0.8
0.6
0.4
0.2
1
1.5
2
2.5
3
3.5
V eloc ity (m m /s )
4
4.5
5
Figure 3.23: The ratio of the transition times for the fluid from 720C to 950C, 950C to
550C and 550C to 720C in the 720C, 950C and 550C as function of mean
velocity
47
Temperature a the wall and in the center of the channe for fluid velocity (4 mm/s)
370
365
360
Wall of the channel
Center of the channel
Temperature(K)
355
350
345
340
335
330
325
0
0.01
0.02
0.03
0.04
Channel Length (m)
0.05
0.06
Figure 3.24: The comparison between the temperature at the channel wall and center for a
velocity of 4 mm/s.
3.17 CONCLUSION OF FINITE ELEMENT SIMULATION
Simulation results show that single-sided heating produced an adequate temperature
profile but has longer thermal transition lengths. Two sided heating would increase
complexity of the fabrication process with the payoff of more uniform temperature zones
while single side heating reduced complexity. Operating the chip at velocities of higher
magnitude would increase thermal power requirement. This leads to an increase in the
temperature just beneath the heaters, which may get too close to the polycarbonate glass
48
transition temperature, therefore two side heating would be more favorable at velocities
greater than 2 mm/s.
The mean velocity in the channel was increased from 0 mm/s to 5 mm/s. For every
increase in the fluid velocity, more convective heat was removed from the 950C and the
720C zones so the constraint temperature tended to fall. However, heat was transferred
from the 950C to the 550C zone, which resulted in a temperature increase. For higher
velocity the fluid spends more time in transition, than in the constant temperature zones.
The maximum recommended mean velocity for this device is 4 mm/s (maximum velocity
of 7.5 mm/s). The results of comparison between the temperature at the channel wall and
center showed a difference of 1.3 0C between the maximum temperatures.
49
CHAPTER 4: FABRICATION OF THE MOLD INSERT BOTH PCR DEVICES
4.1 INTRODUCTION to (LIGA) MOLD INSERT
Mold inserts were fabricated using X- ray LIGA microfabrication techniques for the
test (straight channel) [Bejat, 2001] and the (final spiral) devices. Initially, UV-LIGA was
used to fabricate a mold insert for the spiral design configuration but the height was
limited to 100 µm [Liu, 2003]. Deep X-ray lithography permits fabrication of
microstructures with aspect ratios greater than 100:1 and height over 1 mm.
X-ray LIGA was used because of the ability to fabricate high, smooth, vertical
sidewalls at high aspect ratio. The X-ray mask was used to reproduce the desired features
in poly methyl methacrylate (PMMA X-ray resist), which was attached to a stainless steel
plating base. Following X-ray exposure, the exposed PMMA was dissolved in a chemical
developer (GG developer) and Ni was electroplated into the pattern to produce the
desired mold insert structure. The structure was lapped and polished to obtain the desired
final height.
These mold inserts were used to hot embosse high aspect ratio
microstructures of polycarbonate. The embossed polycarbonate devices were then used as
the microfluidic channels.
4.2 BACKGROUND OF LIGA FABRICATIONS
LIGA (an acronym from the German words for lithography, electroplating, and
molding) is a micromachining technology used to produce micro-electromechanical
systems (MEMS) mainly in metals, ceramics or plastic [Kovaces, 1998]. This processes
uses synchrotron X-ray radiation as a lithographic light source. Highly parallel X-rays
from the synchrotron impinge on a mask patterned with high radiation absorbers. The
absorbers on the mask are thick enough to prevent the penetration of X-rays. In the open
50
areas of the mask, the radiation passes through and exposes the photoresist typically poly
methyl methacrylate (PMMA). In most cases the PMMA is attached to the substrate that
is used later as an electroplating base. Bond scissions occur in the region of the PMMA
that is exposed to the X-ray, which are selectively dissolved in a chemical developer.
Once the PMMA is developed, the resulting pattern is filled with metal by
electrodeposition from its conductive base.
LIGA
processing
contains
two
distinct
electroplating
applications.
One
electroplating process is the deposition of gold onto a mask substrate to serve as the Xray absorber during synchrotron exposure. The gold thickness is typically about ten to
twenty microns and must be dense, homogeneous, and uniformly thick to provide an
adequate contrast ratio. The absorber stress must also be minimized otherwise it can
cause pattern displacement errors. The second application of electroplating in LIGA is
the fabrication of the metal device, which requires plating in a pattern of thick X-ray
resist, with lateral dimensions on the order of tens of microns or more, and heights on the
order of millimeters.
4.3 OPTICAL LITHOGRAPHY
The first step was to fabricate an optical lithography mask and transfer the pattern
to an X-ray mask [Desta, 2000]. Optical lithography is the first step in LIGA, where the
design is transferred from paper to a physical entity in the form of an optical mask. The
optical mask consists of a radiation-absorbing component laid out in a pattern based upon
the device design. For ultraviolet (UV) light based optical lithography, 300 Å of
chromium forms the radiation-absorbing region. Dark field regions on the mask are
covered with chrome and block the UV light. Clear field regions allow the UV light to
51
pass through and modify the photoresist. The light either polymerizes a negative resist
into chains or causes molecular scissions in a positive resist. The masks where dark field
regions define the pattern are called dark field masks, while those where the clear filled
areas define the pattern are called clear field masks. Dark field and clear field masks for
the same pattern have chrome on complimentary areas and are in “opposite tone” to each
other.
4.4 OPTICAL MASK DESIGN
An optical lithography mask pattern of a straight channel PCR device was laid out
using AutoCAD 2000 (AutoDesk, San Rafael, CA). The minimum size feature on the
mask was 50 µm, which was the channel width and the channel length was 6.5 cm. The
AutoCAD drawing was converted to the .dxf file format. This .dxf file was then
converted to the .gds file format using the DXF2GDS (DFX, Artwork Conversion, Santa
Cruz, CA) conversion software then viewed using the GDSVUE software. By doing
these conversions of the mask both the dark and clear field images of the mask were
assessed for errors before sending them out for bid. After the assessment, the industry
standard chrome on soda lime glass, 0.09” thick and each side, 5” by 5”, optical masks
were purchased from LSI Photomask (Scottsdale, AZ).
The mask, shown in Figure 4.1 (b) consists of straight micro fluidic channel in the
center length 6.5 cm and width 50 µm. There were 20 other channels on either side of the
fluidic channel. They 300 µm by 4 cm and were used to support in section of the
temperature-sensing devices (thermocouples) for the control circuit assembly. At each
end there was a reservoir; these were used as input and output ports for fluidic
interconnects.
52
Enlarged Section
Micro fluid channel
Reservoir
Thermocouple channel
Figure 4.1(a): Optical mask pattern for the PCR straight channel designs
Figure 4.1(b): Optical mask purchased from LSI Photomask
(Scottsdale, AZ).
For the spiral design, Yannick Bejat laid out the mask, which was also purchased
from LSI Photomask (Scottsdale, AZ). The mask for the spiral device can be seen in [Liu,
2003] or in Figure 4.2.
53
Figure 4.2: Optical mask pattern for the PCR Spiral device conjuration designs by
Yannick Bejat [Bejat, 2001]
4.5 X –RAY LITHOGRAPHY
X-ray lithography uses synchrotron X-ray radiation as a lithographic light source.
The highly parallel X-rays from the synchrotron impinge on a mask patterned with X-ray
high radiation absorbers. The absorbers on the mask are thick enough to prevent the
penetration of X-rays. Hard X-rays are used as the exposure radiation. The synchrotron
ring at the Center for Advanced Microstructures and Devices (CAMD) at Louisiana State
University was used for this application. This is an electron storage ring with electron
storage energies of 1.3 GeV and 1.5 GeV. Due to the use of shorter wavelengths and the
insignificance of diffraction, X-ray lithography is far superior to optical lithography for
54
producing aspect ratios in the order of 20:1 or higher. The use of PMMA as an X-ray
resist has been well documented [Pan, L.W et. al, 2001] [Madou 1997]. The exposure and
development of PMMA resists can be done as long as the X-ray source provides 4 to 25
KJ/cm3 of energy for a maximum top to bottom dose ratio of 5 [Madou 1997]. The
exposure dose was carefully calculated using Transmit [http://www.camd.lsu.edu,
2000]. Overexposure can cause swelling and cracking which may result in deformation or
delamination of the desired structures. Underexposure may cause PMMA to be
undeveloped, resulting in the inability to electroplate metal on the surface of the
substrate.
4.6 X-RAY MASK FABRICATION TECHNIQUES
The mask for X-ray lithography mask consists of a substrate patterned with a high
radiation absorber. The substrate must be adequately transparent and the absorber opaque
to the incident radiation. Some of the commonly used mask substrate materials are
Silicon (Si), Silicon Carbide (SiC), Silicon Nitride (SixNy), Beryllium (Be), Titanium (Ti)
and Graphite (C) (Madou, 1997). For this application graphite was used as the substrate
for the mask. Graphite also acts as a filter for X-rays of shorter wavelength, thus
transmitting a specific bandwidth of X-ray radiation to the PMMA. High radiation
absorption is proportional to molecular weight cubed. Gold, with a high molecular weight
and density (Mwt=196, ρ=19300 kg/m3), with an absorption coefficient, α, ranging from
0.8 – 11 µm/m (Madou, 1997), and can be electrodeposited was used for this application.
The X-ray mask was constructed directly on a graphite membrane using SU-8 25
negative resist (MicroChem, Newton MA). For a negative resist that the exposed areas of
the resist cross-link into long chains. Prior to resist application graphite wafers [Poco
55
Decatur, TX, DFP-3] were polished and cleaned thoroughly using acetone, IPA, and DI
water to remove all unwanted particles from the wafer surface. To avoid sagging of the
graphite wafer under its unsupported weight plus the weight of the resist, the wafer was
bonded to a standard NIST (National Institute of Standards and Technology) aluminum
adapter ring after processing the resist and gold electrodeposition. Sagging generally
causes underexposure because of the gap change in the between the mask and the wafer.
This generally leads to resist being under developed during resist development.
4.7 SPIN COATING AND PRE-BAKE OF RESIST
The backside of the graphite wafer was placed on a spin coating machine, and
fixed into place via a vacuum chuck. Approximately 15 ml of SU-8 25 (MicroChem
Corp., Newton, MA) negative photoresist was placed on the wafer and spun at 1000 rpm
for 20 seconds to give a uniform coat of the resist. A thickness of 50-60 µm of resist was
obtained on the graphite wafer. The SU-8 coat was pre-baked in a convection oven,
starting at a temperature of 60°C, then ramped up to 95°C at a rate of 4°C/min and
maintained at this temperature for about 1 hour, then slowly cooled down to 30°C at
2°C/min in the oven. Final cooling from 300C to room temperature (~230C), took place
outside the oven. The slow cooling was necessary to reduce the internal stresses
generated in the SU-8 layer during solidification.
4.8 UV EXPOSURE, POST BAKE AND DEVELOPMENT OF RESIST
The SU-8 coated wafers were exposed to ultraviolet light through the optical mask
for 18.3 seconds. The wafers were exposed to 330 mJ/cm2 of UV light, which was the
required energy for exposing a 50 µm thick layer of SU-8. After exposure the wafers
were post-baked in a convection oven, starting at a temperature of 60°C, ramped up to
56
95°C at a rate of 4°C/min and maintained there for 15 minutes, after which they were
cooled down to 70°C at 1°C/min. Cooling to room temperature was done in ambient air.
The post-baked, patterned wafer was developed in the SU-8 developer
(MicroChem Corp., Newton, MA) for a total of 6-8 min, moving back and forth from a
cleaning bath to a rinse in IPA in 3 steps. The appearance of a white residue indicated
incomplete development of SU-8. Whenever there was any white residue the wafers were
re-immersed in the developing solution for a few more seconds. After complete
development they were rinsed and dried with compressed air.
4.9 GRAPHITE SUBSTRATE PREPARATIONS AND GOLD PLATING
A layer of gold was electroplated in the developed pattern on the graphite wafer
to produce the absorber layer for the X-ray masks. A commercially available neutral gold
sulfite solution (Technic Inc, Irving TX), containing 1 troy oz gold per gallon was used
for gold plating. Before plating 7.5g/l of sodium sulphate (Na2So3) was added to the gold
solution. The gold sulfite (Na3Au(SO3)2) solution, was mildly acidic, with a pH of 6 – 7.
Plating was carried out at a current density of 5 mA/cm2, using a pulsed current (1 Hz)
for four hours to get a 12-15 µm coat of gold. The plating rate was approximately 4 µm
per hour. A platinized titanium mesh was used as the anode. The bath temperature was
maintained 25°C and the amount of gold in the solution was maintained at no less than
80% of its initial value. This was done by more concentrated commercially available
neutral gold sulfite solution (Technic Inc, Irving TX), containing 1 troy oz gold per
500ml. The amount of gold used in plating was from the gold sulfite (Na3Au(SO3)2) bath
was 1 troy oz gold for every 4 mA/hr.
57
Figure 4.3: Fabricated X-ray mask for PMMA resist exposure.
The structure of the mask was composed of SU-8 and with gold embedded in it.
The exposed SU-8 resist was left on the wafer to provide structural stability to the mask.
The mask was now ready for use. It was attached using super glue (Loctite 4210, Loctite
Corporation, Newington, CT) to an NIST (National Institute of Standards and
Technology) aluminum adapter ring mask is shown in Figure 4.3. The striations in the
gold are due to secondary roughness arising from the graphite surface.
4.10 PREPARATION OF STAINLESS STEEL MOLD INSERT SUBSTRATE
In order to keep costs low for individual nickel electroforms, it was necessary to
find an inexpensive substrate, which the raised microstructures could be plated upon.
Other properties such as machinability and thermal the coefficient of expansion were
58
considered as well. Good machinability was necessary in order to easily cut the substrate
to the desired dimensions and obtain an initial surface finish to rms value of 1µ. Finally,
thermal expansion characteristics had to be considered because nickel microstructures
electroformed directly onto a substrate with a large difference in coefficient of thermal
expansion could debond under the elevated temperatures used during embossing. The
coefficient of thermal expansion of 316 stainless steel (16 µm/m-°C) and nickel (13
µm/m-°C) are very similar.
Therefore, stainless steel was judged as an acceptable
substrate for the present application.
The 316 stainless steel (Ryerson-Tull, Dallas, TX) was cut and machined into 5inch diameter base plates. Using a Harig 618 surface grinder (Elgin, IL), the base plate
was ground flat and sandblasted to create a roughened surface with rms value of
approximately 2 µ. The stainless steel base plate was activated in C-12 activator (Puma
Chemical, Warne, NC) with a current density of (20 mA/cm2), for 2 minutes forward and
2 minutes backward. Approximately 10 ml of a 9% PMMA in cyclobenzene solution
(MicroChem, MA, USA) was placed on the wafer and spun onto the base at 1000 rpm for
about 30 seconds. This was allowed to melt on the stainless steel surface at 180 0C in a
conventional oven and cooled slowly to room temperature.
4.11 PMMA BONDING
The resist, 500 µm thick PMMA sheet stock (CQ grade Goodfellow, UK) was cut
to the size of the substrate and bonded to its coated surface. A solvent bonding process,
using a MMA bonding solution was used to achieve the bond. The constituents of the
bonding solution are shown in Table 4.1
59
In order to mix the solution, the powdered PMMA was dissolved in the MMA. This
process took a fairly long time during which some of the MMA may evaporate. It was
kept in a closed container and the weight of the container was recorded before and after
the PMMA dissolved in the MMA so that the correct amount of MMA could be added to
maintain the correct proportions if any had escaped from the closed container. The DMA
acted as a starter for polymerization and BPO was used as the hardener. The BPO was
not added until a few minutes before use. After adding BPO the mixture hardened within
15 minutes. The above solution was thoroughly mixed and degassed in a fume hood for
12 hours.
A few drops of this MMA glue were applied to the wafer surface and the 500 µm
thick PMMA resist was pressed onto the substrate. Care was taken to ensure that no
bubbles were trapped in the glue. This sandwiched wafer was placed in a pneumatic press
for 12 hrs under a pressure of 10-12 psi at room temperature. Finally, the PMMA was fly
cut to the final desired height of the microstructures, which in this case was 200 µm,
which leaves us with 50 µm for surface grinding and polishing. The thickness of the
PMMA resist needed to be known to enable accurate calculation of the dose for X-ray
exposure.
Table 4.1 Composition by mass of MMA bonding glue
MMA (Methyl methacrylate)
8.5 gm
Powdered PMMA
BPO (Benzoyl peroxide)
1.5 gm or 0.50
0.15g
DMA (Dimethyl aniline)
0.1g
MEMO (Methacryloxpropylltrinethoxysilane)
0.1g
60
4.12 X-RAY EXPOSURE OF PMMA
To pattern structures in PMMA accurately by the shadow printing process, a high
energy, highly collimated light source was required. A synchrotron provides such a light
source. A synchrotron is an electron storage ring where relativistic particles travel at
nearly the speed of light, emitting a continuous spectrum from infrared to the hard X-ray
range.
X-ray exposures were performed on the XRLM-3 beamline at CAMD. A 125-µm
Beryllium filter was installed in the beamline to act as an initial filter for low energy Xrays. The intensity of X-rays decreases as they pass through the different filters, and
mask. The exposure dose must be calculated as shown in Appendix 4.A. Overexposure
can cause swelling and cracking resulting in deformation of desired structures, while
underexposure results in the inability to electroplate a metal at the substrate surface. For
full exposure of PMMA, the bottom surface of the resist must receive at least 3000
mJ/cm3 of X-ray energy and the top surface should be less the 20000 mJ/cm3. Thicker
samples will require longer exposure times in order to deposit the necessary 3000 mJ/cm3
at the bottom. In this instance, the thickness of PMMA to be exposed was 200 µm.
Combinations of 6 µm and 15 µm thick aluminum filters were used to improve the
contrast between the top and bottom surfaces of the resist. The filters also served as an
additional barrier for low energy X-rays, thus improving resolution. The more filtration
used, the better the contrast between the exposed and the unexposed resist, although the
time for complete exposure can increase significantly.
61
4.13 PMMA RESIST DEVELOPMENT
X-ray radiation causes scissioning of the polymer chains. The broken chains have
lower molecular weight and can be dissolved in a molecular weight-sensitive organic
solvent. To fully utilize the accuracy potential of synchrotron radiation lithography the
resist-developer system should have a ratio of dissolution rate in the exposed and
unexposed areas of approximately 1000. This would result in minimal dissolution of the
unexposed, high molecular weight cross-linked PMMA and a sufficient dissolution rate
in the exposed area. See Table 4.2 for the composition by volume of the GG developer
and GG rinse used to dissolved exposed PMMA
Table 4.2: Composition by volume of the GG Developer formulated at the Institute for
Microstructure Technology at the Forschungszentrum, Karlsruhe. This solvent
is used to dissolve PMMA that were exposed to X-rays.
(ml)
(ml)
Developer:
2-(2-Butoxyethoxy) ethanol / (diethylene glycol
1500
900
Morpholine
500
300
2-aminoethanol
125
75
Water
375
225
2500
1500
2000
1200
500
300
2500
1500
butyl ether)
Total:
Rinse
2-(2-Butoxyethoxy) ethanol / (diethylene glycol
butyl ether)
Water
Total:
62
For every 200 µm of PMMA thickness, the wafer was run through a development
cycle of 20 min in the GG developer and a rinse cycle of 40 min in the GG rinse at room
temperature for three repetitions. The GG developer was filtered continuously. After
development the microstructures were rinsed with deionized water and dried with
compressed air before electrodeposition.
4.14 NICKEL ELECTRODEPOSITION
In order to achieve a good bond between the plated nickel and the stainless steel
substrate, the substrate was activated with C-12 and Wood’s strike before nickel was
electrodeposited. A nickel sulfamate electroplating solution {see Table 4.3} was used to
electrodeposit nickel into the exposed and developed PMMA mold.
Before plating the stainless steel base plate with the PMMA patterned was
activated again using a C-12 activator to remove any dirt or residual oxide prior to
electroplating. The C-12 bath was composed of 35 of gallons water and 5 of gallons of C12 activator with 2% HCL to reach a pH of 1.5. By using nickel pellets as the anode and
the work piece as the cathode, the substrate was activated with a current density of (20
mA/cm2) at room temperature. This was done with cycle of two minutes forward and two
minutes backward.
The substrate was treated with the Wood’s strike because the chromium atoms in
stainless steel have a tendency to form an oxide layer across the surface of the substrate
[Stefanov, 2000]. This oxide layer creates a weak bond between the substrate and the
electroformed nickel. Debonding is a significant problem when structures are plated
directly onto this oxide layer. In order to minimize this problem, an acid Wood’s strike
composed of 240g of nickel chloride (NiCl2*6H2O), 80g of hydrochloric acid (HCl) and
63
Table 4.3: Composition by mass of the nickel sulfamate solution
Component
Chemical formula
Amount (g)/liter
Nickel sulfamate
Ni (SO3NH2)2.4H2O
450
Boric acid
H3BO3
37.5
Sodium lauryl sulfate
C12H12O4Sna
1.0
one liter of distilled water at a pH of 1.5 was used. This was done at room temperature
(23 0C) with reverse plating for two minutes at a current density of 50 mA/cm2, then
switched to forward plating for six minutes at 50 mA/cm2. The nickel strike had two
effects on the stainless steel substrate. First, the acidic nature of the bath etched away the
oxide layer. Secondly, a thin layer (~ 1 µm) of nickel was plated onto the oxide free
surface giving excellent adhesion between the stainless steel substrate and the plated
microstructures. The substrate was then placed directly into the nickel-plating bath
without rinsing.
Nickel sulfamate electroplating solution was used to electrodeposit nickel into the
exposed and developed PMMA molds directly after the Wood’s strike. Nickel
electrodeposition using a nickel sulfamate solution has been widely used for
electroplating into microstructures (Madou, 1997). Nickel sulfamate has the advantages
of low stress in electroplated microstructures, as compared to a nickel sulfate or nickel
chloride electroplating solutions.
Six liters of 1.8 M nickel sulfamate electroplating solution was prepared by using
the compounds listed in Table 4.3. The concentration of the hydroxyl ions (OH-)
increases as the plating progresses and increases the pH of the solution. A starting pH of
64
3.8 was used to account for the increase in pH during electroplating. Boric acid was used
as a buffering agent to minimize the pH rise of the solution while electroplating. Lauryl
sulfate served as a wetting agent reducing the surface tension of the solution, which
helped the solution to reach deep recesses in microstructures and also contributed to low
stress plating. Also plating at lowering current densities can reduce stress, by reducing
grain sizes of the plated nickel. Plating nickel to a height of 200 µm, to allow for post
plating finishing required only 10-12 hours using a current density of 25 mA/cm2.
The components were mixed and DI water was added to increase the volume to six
liters. The solution was heated on a hot plate, with constant stirring, to a temperature of
about 50-60 °C. It required about two hours to mix and dissolve the salt completely. After
mixing, the nickel sulfamate plating bath was filtered with a 0.45-um-celluose filter prior
to electrodeposition.
4.15 LAPPING AND POLISHING
The final step in producing the mold insert was lapping or surface grinding of the
substrate to the final thickness and then polishing. This was done using diamond slurries
with an ENGIS lapping machine (Model 15LM11V Engis Wheeling, IL). Thickness
tolerances were plus or minus five microns, and the surface finish was one-micron rms.
The surface-grinding step was used to ensure that the surfaces were parallel to each other
was able to produce a flatness of under 5 µm across the 5 inches diameter of the stainless
steel substrate. Once ground parallel to the bottom of the substrate at the desired height,
the tops of the electroformed structures were polished using diamond coated sand paper.
The smoother the surface the easier is demolding.
65
Finally, all of the PMMA on the substrate was flood exposed and developed in the
GG developer. The mold insert was then complete and ready for hot embossing.
Both mold inserts were inspected and scanning electron microscope images of them
were taken using the SEM and Electron Microprobe Laboratory at the Department of
Geology and Geophysics at LSU. Figure 4.4 shows the SEM image of the straight
channel mold insert and Figure 4.6 shows the SEM image for spiral design mold insert.
Figure 4.5 and 4.7 are actual photos of the straight channel and the spiral design mold
inserts respectively.
Stee Substrate
Nickel Plated
structures
Figure 4.4: SEM of straight channel mold insert
66
Figure 4.5: Actual photo of the straight channel mold insert
Nickel Plated
structures
Stee Substrate
Figure 4.6: SEM of spiral channel mold insert
67
Figure 4.7: Actual photo of the spiral design mold insert
68
CHAPTER 5: THERMAL SYSTEM DYNAMICS AND CONTROL
5.1 INTRODUCTION TO SYSTEM DYNAMICS OF THE PCR DEVICE
The system dynamics of the PCR device were studied to determine: (1) the required
thermal input for the selection or design of the heater; (2) to determine the controller
needed to maintained the steady state temperatures for the PCR cycle and (3) to check the
stability and time constant.
One of the design criteria for the PCR was to achieve one-dimensional heat transfer,
which required making the device as thin as possible. The heater that would be selected
or designed needed to meet several specifications to satisfy the assumption of onedimensional heat flow, including: (1) thin < 50 µm so that heat flow in the lateral
directions could be neglected, (2) capable of proved the required power for each constant
temperature zone. The area of the heater was also restricted to cover the spiral path of the
CFPCR prototype.
5.2 PSEUDO BOND GRAPH MODELING FOR THERMAL SYSTEMS
In order to study the dynamics of the system a lumped parameter thermal model of
the PCR device was developed using the pseudo-bond graph method [Karnopp, 2000].
Thermal systems have no inertial element so that the only one-port elements that were
used in this system model were sources, resistors and capacitors. In the case of thermal
systems, the product of temperature and heat flow is not equal to power. Heat flow itself
has the dimensions of power. For this reason, a thermal bond graph with temperature as
the effort variable and heat flow as the flow variable is called a pseudo bond graph.
The PCR was assumed to be symmetric with the same power input on either side,
which was modeled as a constant heat flux (flow source). The ambient condition was
69
taken as constant temperature, T∞ , which was represented by an effort source. Since the
ratio of the thickness compared to the length and width of the device was less than 0.001,
one-dimensional heat transfer was assumed.
The PCR was reticulated into three sub-systems capable of storing energy; these
represented the air in the air pockets, the polycarbonate cover plate, and the core of the
system, which was comprised of water (to simulate the analyte) and polycarbonate. The
capacitances were evaluated using Equation 5.1.
C = mc p
5.1
where c p was the heat capacity at constant pressure, and m was the mass. Upon
evaluation, the capacitance of air was four orders of magnitude less than all the other
capacitances as shown in Table 5.1 and energy storage in the air was neglected.
Heat transfer to the environment was assumed to be by convection, and the
resistance was represented by (Equation 5.2):
Rconv =
L
hc Asurf
5.2
where hc is the local convective heat transfer coefficient, and Asurf is the surface
area.
Table 5.1: Numerical values for the resistors and capacitors for the lumped parameter
model
R1=R5
R2= R4
R3
C1= C3
C2
Cair
32 m2K/W
114.8 m2K/W
14.55m2K/W
3.8 J/k
1.5155 J/k
1.59x10-5 J/k
70
Heat flow between the sub-systems was modeled by resistive elements for conduction
(Equation 5.3) and radiation (Equation 5.4).
Rcond =
L
kA
5.3
where L is a characteristic length, k is the thermal conductivity of the material, and A is
the cross-sectional area. Radiation is a nonlinear function of temperature, so the
relationship was linearized by assuming that the temperature difference between any two
surfaces was small when compared to the absolute temperatures of the two surfaces. A
3
)
linearized radiation heat transfer coefficient was calculated ( hr = 4σTsurf
R rad =
1
hr Asurf
5.4
The total resistance across the air pocket was:
R total =
LT
A(k Th + hr L T )
5.5
as shown in Section 3.4, where both the radiation resistance and the conduction resistor
through the air were parallel to each other. After simplification, the model was reduced to
three capacitances and five resistances. The electrical analogy was used to layout the
equivalent circuit diagram and a bond graph was drawn for the system as shown in Figure
5.1 and 5.2 respectively.
71
R2
R1
R5
R4
C1
C2
I
V0
I
R3
V0
C3
Figure 5.1: Equivalent electrical circuit for one dimensional heat flow for each zone
Se1
Se
R1
C1
R2
C2
R5
C3
R4
R
C
R
C
R
C
R
1
0
1
0
1
0
1
Sf
Sf1
R
Se
Se2
Sf
R3
Sf2
Figure 5.2: Pesude bond graph of the system
The linearized state equations were rearranged in matrix format [Equation 5.6]
dT1 
 dt 
 
dT2  =
 dt 
 
dT3 
 dt 
− ( R1 + R2 )
T1
C1 R1 R2
1
R2 C 2
0
1
R1C1
−1 1 1 1 


C 2  R3 R2 R 4 
1
R4 C 3
1
R1C1
0
1
R4 C 2
T1 
T 
 2+
T3 
− ( R 4 + R5 )
C 3 R 4 R5
72
0
0
2
C2
0
1
R4 C 3
T∞ 
ei 
 
5.6
5.3 CONTROL SYSTEMS BACKGROUND
In a regulated system, the setpoint specifies the desired response level at the system
output. In this case the output is temperature. In a perfect world, you would provide the
setpoint signal as the input to a system and it would immediately provide the desired
output level. This configuration is called open loop control and is illustrated in Figure
5.3.
A real process does not normally operate in this way. The output may drift away
from the setpoint for a variety of reasons, or may respond too slowly to changes, or may
oscillate too much in response to transient changes. In these situations, it is necessary that
the output of the system be measured and processed by a separate controller (sometimes
called a regulator). The controller compares the system output to the setpoint level to
determine a new control input for the system. Often the difference between the output of
the process under control and the reference input is amplified and used to control the
process so the difference is continually reduced. This configuration, called a closed loop
feedback control system, is illustrated in Figure 5.4.
Closed loop control handles
disturbances better and the time constant is adjustable.
Setpoint
Desired
Response Level
Ideal system
Figure 5.3. Ideal system with open loop control
73
Setpoint
Controller
Desired
Response Level
System
Figure 5.4: Typical system under closed loop control.
Step Response
420
400
1.415 W minimum pow er
2.5 W maximum pow er
Temperature (K)
380
360
340
320
300
0
100
200
300
400
500
600
Time (sec)
Figure 5.5: The response of the system for a step input power of 1.415 W and 2.5 W
which are flow source.
74
700
The open loop step response of the system was evaluated using MATLABTM (ver.
5.4, The Mathworks, Natick, MA) for the power input needed to reach the maximum
required steady state temperature of the device (95 0C) and the glass transition
temperature of polycarbonate (148
0
C), which are plotted in Figure 5.5. These
temperature achieved by constantly adjusting the power input until the desired
temperatures were reached
5.4 STABILITY
A system is stable if all the poles of the function have negative real parts [Franklin,
et al., 1994]. The poles of the system were -0.0510, -0.088 and -0.0167, which were all
negative and real so the time response will decay.
5.5 HEATER POWER REQUIREMENT
For the open loop analysis, of the double side-heating configuration, 1.415W of
thermal power was needed obtained the maximum working temperature (950C) and 2.5W
for the glass transition temperature (1480C) of PCR device. The assumed power inputs
for the single side-heating configuration were approximately 3W for (950C) and 5W for
(1480C) respectively. The average of these values (4W) was chosen as the required power
rating for the heaters. The time constant for the system is 48 sec.
5.6 PID CONTROLLER PARAMETERS
The power (W) from the PID controller is:
d


W (t ) = K P ×  (Ts − To ) + K D (Ts − To ) + K I × ∫ (Ts − To )dt 
dt


5.7
where Ts and To are the final and initial temperature of the system, respectively.
Multipliers KP, KI and KD are referred to as the proportional, integral and the derivative
75
gains. The effects of each of the controller gains on a closed-loop system are summarized
in Table 5.2. MATLAB (SIMULINK) was used to determine the required gains and the
block diagram used is shown in Figure 5.6.
Selecting optimal values of KP, KI and KD parameters of a closed loop control
system is an iterative process. This process is called PID tuning. The MATLAB Controls
Toolbox was used to simulate the time response of the PCR device. The time response for
each zone is shown in Figure 5.7.
Table 5.2: The effects of each of the controller gain on a closed-loop system
CLOSED LOOP
RISE TIME
OVERSHOOT
SETTLING TIME
STEADY-STATE
RESPONSE
ERROR
KP
Decrease
Increase
Small change
Decrease
KI
Decrease
Increase
Increase
Eliminate
KD
Small change
Decrease
Decrease
Small change
0
u(t)
Step
0.01
2.639s2+0.1517s+0.002097
s
s3+0.2362s2+0.008565s+7.094e-005
Integral
Process Transfer Function
Disturbance
y(t)
.8
68
Proportional
Constant
du/dt
-K-
Derivative
Figure 5.6: The block diagram used in SIMULINK to determine the correct gains for the
simulation
76
Step Response
370
360
368 K Zone
345 K Zone
328 K Zone
Temperature (K)
350
340
330
320
310
300
0
50
100
150
200
250
300
350
400
450
500
Time (sec)
Figure 5.7: The transient temperature response for each zone with PID control
The results from the simulation show that 200sec after the system was powered up
the PCR process could be started. If also shows that the temperature as measured by the
thermocouple, in each zone can be held constant using a PID controller with the correct
gains. Three Series 96 PID controllers were bought from Watlow (Winona, Minnesota).
They were capable of self-tuning. Self-tuning allowed the controller to explore the
responsiveness of the system to determine an effective set of gains for PID control. The
controllers could attain PID gains over the full range of gains selected based on the
simulation results so the PCR temperatures could be maintained within the desired babds.
77
5.7 HEATER SELECTION
The next step was to find an off-the-shelf heater that could deliver the 4W defined
by the open loop analysis. Alternatively, the heaters could be designed and fabricated.
An extensive search was carried out for commercially available heaters with the desired
characteristics shown in Table 5.2. The heaters that were available were deposited on,
Kapton or silicone rubber with maximum temperatures of 200°C and 235°C respectively.
The Kapton polyimide heater (Model K005020C5, Watlow, Dallas, TX) was chosen
because it was the only one that was small (2.54 cm x 2.54 cm) and relatively thin
(~200µm) to meet specifications. These heaters were still 150 µm thicker than desired,
therefore it would be best to deposit the heating element directly onto the device.
Table 5.3: The comparison of the specifications for the electrical heaters to commercially
available devices from Watlow (Dallas, TX).
Desire Specifications
Length (cm)
Width (cm)
Power (W)
1.5
Thickness
(µm)
10
Spiral Configuration
2.54
Straight Channel
Configuration
Commercially
Available
Spiral Configuration
1.00
1.00
10
4
Length (cm)
Width (cm)
Power (W)
2.54
2.45
Thickness
(µm)
200
5
Straight Channel
Configuration
3.5
1.2
200
5
78
4
5.8 INTRODUCTION TO HEATER DESIGN
Ideally the heating element will be deposited directly onto the polycarbonate by
activating the surface. Exposing the polycarbonate substrate to UV radiation via a
patterned quartz mask will carry out an activation process as shown by [Henry and
McCarley, 2001] on PMMA.
Electroless plating was used to deposit gold on the
activated surface. After the gold was deposited on the surface, electroless plating was
used to deposit copper onto the patterned gold. The final heating element, a Nichrome
alloy of composition 80% Ni and 20% Cr will be electroplated onto the electroless
copper.
The dimensions of the heating element were determined based on the analytical and
simulation analysis of the heating element. A quartz mask was laid out (see Figure 5.8) so
that the heater could be patterned by UV lithography. The current required to power the
heater was determined.
5.9 RESISTIVE HEATING
Every current-carrying conductor loses electrical energy by I2R losses; this rate of
energy loss is the electrical power that is converted into heat. When a heater is connected
to an electric power supply, current flows through the high resistance heating element.
The heating element resists the flow of electric current through it, converting electrical
energy into heat energy.
The resistance of a conductor of length (L) and cross-sectional area (A) normal to
the direction of current flow is
RE =
ρRL
5.8
A
79
where ρR is the specific resistivity (Ω-m). For metals, specific resistivity increases
approximately linearly with temperature over moderate temperature ranges (20 0C to
5000C). This variation is most often written as
R T1 = R T0 (1 + α 0 (T1 − T0 ) )
5. 9
where αo is the temperature coefficient of resistivity and has units of inverse temperature
(T-1). Because the maximum temperature on the PCR chip is only 95 0C the minor change
in resistance can be disregarded, hence the specific resistivity (Ω-m) of the heating
element was assumed constant. The length and cross-sectional area of the heating
element depends upon the specific resistivity of material and the required power density.
Equation (5. 10) gives the power density, PE for a wire of resistance RE with a voltage VE
applied across its ends or a current IE flowing through it.
The power density for a heater is the ratio of the heat generated by the heating
elements to the surface area of the heater as in Equation 5.11.
(Power Density (PE))
I E2 R E
Watt
=
m2
A surf
5.10
The temperature rise in a material is a function of: (1) The bulk resistance, which
depends on the resistivity and thermal conductivity; (2) the magnitude of the electrical
current; and (3) the convection and radiation energy loss to the surroundings. The
maximum, steady state temperature rise of a heating element in still air was calculated
using Equation 5.11.
80
∆ T = (T H − T ∞ ) =
I 2t 2
I 2t
+
2 kvA 2
vhA 2
5.11
where ∆T temperature rise in 0C, I current in amps, t element thickness in meters, A the
cross-sectional area of the heating element, ν electrical conductivity (1/ρ), k the thermal
conductivity (Wm-1K-1) and h is the natural convective heat transfer coefficient (W/m-2
K-1). This equation gave a conservative estimate of temperature rise because the heat loss
via radiation was neglected [Kakac, et al., 1993].
For the resistive heating applications, the most commonly used heating material is a
nickel-chromium alloy (about 80% nickel and 20% chromium), which carries the trade
name "Nichrome". Nichrome is the most commonly used metallic alloys in resistors used
at elevated temperatures..
All heaters eventually fail or burn out, even when using Nichrome as the heating
element. Resistive heaters will often fail if the heating element is damaged or
contaminated with chemicals, but the primary cause of failure is oxidation. At high
temperatures, oxygen from the air chemically reacts with the element, which causes a
hard oxide coating to form on it. This oxidation process continues until the heating
element eventually disintegrates or burns out. If the heater is maintained at a constant
temperature, the hard oxide coating actually slows the oxidation process and extends the
life of the heater.
Periodic heating and cooling cycles of the heater may cause the resistance element
to expand and contract. This will cause cracking of the hard oxide coating and expose
more metal to oxygen attack. Therefore temperature cycling through abrupt temperature
changes will greatly accelerate the oxidation process and cause the heating element to
81
burn through much more quickly. Even though there is no clear cut way to really stop this
oxidation process, it can be easily slowed down.
One way this was done was to use PID controllers to minimize the temperature
cycling of the heating element. Another way of extending the service life of the heater
was to use it at temperatures lower than the maximum. The maximum temperature on the
PCR chip is 950C and the maximum temperature of the Nichrome heating elements is
about 8700C. By using the heater so far below its maximum temperature the heater
element life will be prolonged.
5.10 DESIGN SPECIFICATIONS AND MASK LAYOUT
The maximum power required from each heater was 4 W. The surface area for each
heater was (length 2.54cm and width 1.52 cm as shown in Figure: 5.8) 3.82 cm2, which
was the area of the spiral the needed to be covered by each heater, and a maximum
temperature increase of (∆Tmax) 135 0C. The variable parameters were the length (Lh),
width (Wh), and thickness (th) of the heating element. Equation 5.13 gives the length of
the heating element, as a function of the system specifications.
2
Lh =
VE A
ρ R PE
5.12
The dimensions chosen for the heater were width Wh: 200 µm, thickness th: 10 µm
and Equation 5.12 was used to calculate the length Lh: 0.29 m. The 10 µm thickness was
chosen because it was the maximum thickness of Ni-Cr that can be plated with direct
current, without cracking (K-Lin, C-Hsu, I-Hsu, T-Chang, 1992). For a heating of cross
section 200 µm by 50 µm, using Equation 5.13 the calculated element length (Lh) was
0.29 m for the required power of 4W and an applied voltage of 25V. Table 5.4 consist of
the electrical properties of commercial Ni-Chrome alloy [Goodfellow, UK]
82
Table 5.4: Electrical Properties of Ni-Chrome alloy [Goodfellow, UK]
Metal
Ni-Chrome
Specific resistivity
ρ0 at 250C (µΩ-cm)
Coefficient
resistivity α(0C-1)
108
0.0004
of Thermal
Conductivity
(W/m-K)
13.4
The predicted resistance of this heater was 156Ω as shown in Figure 5.9 at a current
flow 0.160 amps. The number of turns of the heater coil ( n ) was:
n =
Lh
∆l
5.13
where dl (0.015m) was the length of one turn of the coil. For the above dimensions the
number of turns ( n ) was 20. For the actual design (see Figure 5.8) 24 turns were used to
make allowance for changes in design parameters during the fabrication process, such as
the cross sectional area due to overplating, or the combinations of Nichrome, copper and
gold, which mightresult in a lower specific resistivity. By keeping the thermal power
constant the required resistance of the heating element was calculated as function of the
applied voltage. The relationship between voltage and resistance of the heating element
for a constant power of 4W is shown in Figure 5.9.
83
∆l
Figure 5.8: The mask pattern for the Ni-Cr heater with 24 turns all units are in µm
180
Resistance (Ohms)
160
140
120
100
80
60
40
20
0
0
5
10
15
20
25
30
Voltage (V)
Figure 5.9: The variation of heating element (Ni-Cr) resistance with voltage for a power
of 4W
84
ANSYS (5.7, ANSYS, Inc., Houston, PA) was used to simulate the temperature in
still air (air pockets) as a function of current. The convective heat transfer coefficient was
assumed to be 50W/m2K and the ambient air temperature to be 300K. Figure 5.10 shows
the comparison between the theoretical and simulated predictions of the temperature
increase with an increase in current flow through the heating element. Also the heat flux
from the heating element to the surrounding air was simulated and shown in Figure 5.11.
Tem perature vs Current
500
480
Theoretic al
A NS Y S S im ulation
460
Temperature(K)
440
420
400
380
360
340
320
300
0
0.05
0.1
0.15
Current( am p))
0.2
0.25
Figure 5.10: The temperature increase with an increase in current flow through the
heating element Comparison between the Lumped parameter and finite
element results simulated results.
85
Figure 5.11: The heat flux from the heating element to the surrounding air, which is
equivalent to the heaters’ Power density.
The calculated average value for the power density was 96517 W/m2 and the
maximum value obtained from the simulation was 97768 W/m2. The result of the lumped
parameter and finite element models for the temperature variation and the power density
of the heater were within 4% of each other. The error obtained was a result of the
assumption of one dimensional heat transfer for the lumped parameter analysis, which
neglected heat losses along the other two axes. The finite element simulations accounted
for all heat losses, therefore the values obtained from the ANSYS simulation are slightly
less and more accurate. Since all of the design parameters were met, the design is thus
complete and the next step is fabrication.
5.11 DIRECT ELECTRODEPOSITION ON POLYCARBONATE
Schlesinger and his group Schlesinger, [2000] state that it is possible to obtain
patterned or selective deposition of electroless copper by UV radiation of a
nonconductive substrate through a quartz mask. It is possible to achieve electrodeposition
on the irradiated areas by using a negative mask image. This was also demonstrated by
Henry and McCarley [2001] where copper was deposited on PMMA. This process
86
involves cleaning, surface modification, sensitization, and subsequent catalyzing, and
activating.
This procedure was carried out on polycarbonate, where wet chemical deposition
techniques (electroless plating) will be used to deposit copper. Dr. McCarley’s group will
carry out the process in the Department of Chemistry at Louisiana State University. This
process was not completely optimized. Experiments are still being carried out on the
surface modification of polycarbonate. Therefore the complete fabrication process for the
heaters will be completed in the near future.
5.12 ELECTRODEPOSITION OF NI-CR ALLOY
A series of experiments were conducted to determine the correct plating conditions
for Ni-Cr alloy of composition 80%Ni 20%Cr on copper. These conditions were the pH
of the solution, current density, the bath composition, and the plating temperature.
The objective of the Ni-Cr plating was to obtain an alloy composition of 80%(Ni)
and 20%(Cr) that had equivalent properties to the commercially available Ni-Cr alloy
shown in Table 5.4 of section 5.10. It was reported by Lin and Ho, [1991] that, through
the application of DC plating, smooth dense Ni-Cr alloy of up to 20% Cr and 80% Ni
could be obtained. They also stated that percentage composition of Ni to Cr was a
function of temperature, pH, and current densities for the bath composition shown in
Table 5.5. They used a pH of 4 with a current density of 0.005/cm2 and a plating
temperature of 10 0C to obtain Cr content of up to 30%.
87
Table 5.5: The bath composition for Ni-Cr alloy plating
Names
Chromium Chloride
Chemicals Formulas
CrCl3.6H2O
g/l
100
Nickel Chloride
NiCl2.6H2O
30
Ammonium Chloride
NH4Cl
50
Boric Acid
H3BO3
30
Sodium Citrate
C3H4(OH)(COONa)3.2H2O
80
Fermic Acid
HCOOH, 98%
35ml/l
Another group from Taiwan (Lin, et al. 1992) electroplated Ni Cr on steel using the
same bath composition as Lin and Ho. The plating of chromium produced cracks if the
thickness was over 0.5 µm. The main reason for this was the formation of chromium
hydride, which tended to decompose during electroplating. By using pulse-plating Lin, et
al. (1992) were able to obtain the release of hydrogen during the off period before it
reached the critical thickness of 0.5 µm. They were able to achieve smooth crack-free
plating up to 15 µm of up to 65% chromium with a forward pulse of 5 ms and 3 ms at
zero current.
5.13 PLATING EXPERIMENTS
By using the same bath composition as in Table 5.5 a series of depositions were
carried out in a Hull cell shown in Figure 5.12. The pH was held constant at 2.8 so the
two variables were current density and temperature. The Hull Cell achieved the variation
of the current density so the only parameter that was physically adjusted was the
temperature. Equation 5.14 [West, et al. 1991] was used to calculate the current density
88
variation along the working electrode (Cathode) for an average current density of 12.5
mA/cm2 and the results shown in Figure 5.13.
x
L
1 . 273
 x  x 2 
i( x)
x  

1 . 733 − 0 . 763  (1 + 0 . 0659  −    
=
0 . 359 
 L  L  
I avg
L  

x



1 − 
L


Counterelectrode
y
Working
Electrode
x
θ
L
0,0
Figure 5.12: The Hull Cell bath in Cartesian coordinates
89
5.14
For a current density of 12.5 mA/cm2 and a plating temperature of 6-8 0C an alloy
composition of 21.2% Cr and 78.8% Ni was achieved. The variation of the Nichrome
percentage composition is shown in Figure 5.14.
45
40
Current density (mA/cm2)
35
30
25
20
15
10
5
4
5
6
7
Cathode length (m)
8
9
Figure 5.13: Hull Cell current distribution along the working electrode with average
current of 12.5mA/cm2
90
10
100
Cr (%)
Ni (%)
90
Ni-Cr Composition (%)
80
70
60
50
40
30
20
10
0
5
10
15
20
25
30
35
40
45
2
Current density (mA/cm )
Figure 5.14: Nichrome composition for the current distribution along the working
electrode with average current density of 12.5-mA/cm2.
Based on the calculated results for plating temperature range of 6-8 0C, an alloy
composition of 21.2 % Cr and 78.8% Ni was achieved. This composition was obtained at
highest current density in the Hull Cell bath, which was the 43.5mA/cm2 with a pH of
2.8.
91
CHAPTER 6: FABRICATION ASSEMBLY AND EXPERIMENT SET UP OF TEST
DEVICE
6.1 INTRODUCTION TO FABRICATION AND ASSEMBLY OF TEST DEVICE
The mold inserts were used to hot embossed several microfluidic channels in
polycarbonate substrates. These substrates were then bonded to 250 µm thick
polycarbonate sheets to form closed channels. The air pockets were machined from 3 mm
thick polycarbonate sheet at depth 1mm as shown in Figure 6.4. These polycarbonate
sheets (with air pockets) were cut to specific dimensions as the cover plates. All the cover
plates were machined at the Chemical Engineering workshop at Louisiana State
University (Baton Rouge La).
The cover plates and the closed micro channels were assembled with heaters and
micro fluidic interconnects. The fluidic interconnects used to connect the PCR device to a
syringe pump (the source of flow).
For the experiment Liquid crystal would be used to display the steady state
temperature profile alone the channel. A camera will then be used to capture the image
displayed by the Liquid crystal. These images will then compared to the industrial chart
to assign values to the different colors displayed by Liquid crystal. The temperature
contours obtained from the Liquid crystal were then compared to those achieved from
ANSYS simulation to ensure that the predictions from the simulations were valid.
6.2 POLYCARBONATE MOLDING
Embossing of the microfluidic channels was performed on a PHI (City of Industry,
CA) press equipped with a machined vacuum chamber.
Polycarbonate sheets from
(Goodfellow, UK)) of thickness 5 mm and 2 mm were cut into 13.456 cm diameter round
92
disks. So that it would be fixed in the vacuum chamber. After washing with isopropanol
and distilled water the polycarbonate plates were dried with compressed air and prebaked in an oven at 80°C overnight for 12 hours to drive off excess moisture and residual
monomer. The rounds were placed in a vacuum fixture and held in place with an
aluminum ring. The mold insert was attached to the opposite side of the vacuum fixture
with screws. The complete fixture was then placed between the two platens of the PHI
press shown in Figure 6.1 with the side of the vacuum fixture containing the molding tool
in contact with the top platen and the polycarbonate disk in contact with the bottom
platen. The press was closed with enough pressure for both sides of the vacuum fixture
to be in contact with the platens.
The press was heated with different temperatures applied to the platens. The upper
platen with the mold insert was heated to 190 0C and the bottom platen with the
polycarbonate disk to 900C. When heated 850 lbs force was applied for 5 minutes and
then the part was demolded.
After molding, the polycarbonate plates were placed
between two glass plates and kept in an oven at 130°C for one half hour, then allowed to
slowly cool at 2 0C /min to room temperature. This annealing process was used to reduce
the residual internal stress in the polycarbonate chips.
SEM photos of the molded polycarbonate chips for both the test device and the
spiral device are shown in Figures 6.2 and 6.3, respectively. The sidewalls of the channel
were straight and smooth but the bottom of the channels is relatively rough because of the
sandblasted surface of the mold insert.
93
Top and bottom
Platens
Vacuum Chamber
Figure 6.1: PHI press equipped with a machined vacuum chamber used for hot
embossing of the microfluidic channels
Micro channel
Thermocouple port
Figure 6.2: SEM of molded the polycarbonate using straight channel mold insert
94
Micro channel
Figure 6.3: SEM of molded the polycarbonate using spiral channel mold insert
6.3 POLYCARBONATE BONDING
A polycarbonate sheet (Goodfellow, UK) of thickness 0.25 mm was cut into a
12.7cm diameter disk and used to seal the microchannels. Before bonding, two 0.25 mm
diameter holes were drilled through the molded disk at the inlet and outlet positions of
the device for the assembly of the fluidic interconnects. The molded disk and its 0.25 mm
disk cover plate were thermally bonded together to form the closed channels.
Before bonding these disks were ultrasonically cleaned in distilled water for 1 hour,
then rinsed thoroughly with distilled water and dried with compressed air. Immediately
before bonding, the two bonding surfaces were coated with isopropanol and dried with
compressed air. The embossed substrate and its cover were clipped between two highly
polished glass plates and placed in a convection oven. The optimized bonding conditions
for the polycarbonate chip were heating for 30 minutes at 150°C, followed by a slow cool
down to room temperature a rate of 5 °C/min. The enclosed micro fluidic channel was
now complete.
95
6.4 FLUID INTERCONNECTS
The two 0.5 mm diameter holes that were drilled in the molded disk were the inlet
and outlet ports for the chip. Polycarbonate female adapters with 5 cm inside diameter
were designed and built for the connection of the fluidic assembly. These adapters were
bonded to the device with epoxy. Fluidic assemblies from Upchurch Scientific (Oak
Harbor, WA) were used to connect capillaries (Polymicro Technologies, Phoenix, AZ) of
20 cm length, 75µm i.d. 375 µm o.d to the device. A syringe pump with a syringe-tocapillary adapter was used to make the connection between the pump and capillary from
the device. This syringe pump (Harvard 22, Harvard Apparatus, Holliston, MA) was used
to pump the analytethrough the microchannel during experiments.
6.5 ASSEMBLY OF THE TEST DEVICE
Commercial thin film, resistance heaters (5 W, 24 VDC, K010030C5, Watlow,
Columbia, MO) were used to provide the heat fluxes to maintain the temperatures in the
isothermal zones of the PCR device. Theses heaters were fixed in their respective
positions with high thermal conductivity element (Omegatherm 201, Omega, Stamford,
CT). The cover plates with air pockets were place directly over their respective heaters
and fixed in position with mechanical clips. Figure 6.4 is a drawing of the cover plate
with air pockets.
96
Air pocket 3
Air pocket 2
Air pocket 1
Figure 6.4: Cover plate with machined air pockets
6.6 CONTROL SYSTEM CIRCUIT ASSEMBLE
The heaters were connected to a control circuit which consist of; (1) computer to
measure on collect data of the transient response of the system, (2) electrical power
supply, which power the heaters, (3) (PID) controllers to control the temperature in each
isothermal zone, (4) temperature sensors (thermocouple) for the closed loop feedback as
shown in Figure 6.5.
The control circuit consists of a power supply, thermocouple, electrical relays, PID
controllers and thin film electrical resistive heaters as shown if the schematic of Figure
6.6. K type thermocouples (HYPO-33-1-T-G-60-SMP-M, Omega, Stamford, CT) were
used for temperature feedback to the PID controllers (Series 96, Watlow, Winona, MN).
97
The controllers were equipped with relays that were used to regulate the current to the
heaters during the temperature control process (see Figure 6.6).
PID Controller
Electrical Relay
(DC) Power
Supply
Thermocouple
Electrical heater
Figure 6.5: A schematic diagram of the control circuit used to control the temperature of
the PCR device.
After the test device was the assembled it was coated with liquid crystals (LC)
(Davis Liquid Crystals, San Leandro, CA). These were used to measure the temperature
distribution along the channel and completely characterize the thermal performance of
the device. LC thermography is a temperature measurement method that uses the colortemperature response of thermochromic liquid crystals (TLCs) to determine the actual
temperature distribution along the PCR channel.
The activation temperature and bandwidth of a TLC formulation are determined by
its chemical composition at the time of manufacture. The TLC activation temperature is
the temperature at which a TLC will begin to reflect visible light. The temperature
98
bandwidth defines the relative color response range for the TLC formulation. TLC
formulations with activation temperatures ranging from -30°C to 120°C and bandwidths
ranging from 0.5°C to 30°C are commercially available, (See Davis Liquid Crystals, San
Leandro, CA) and (Hallcrest, Inc., Glenview, IL) for more detailed information).
Two separate batches of liquid crystals were purchased from Davis Liquid Crystals,
San Leandro, CA of
temperature ranges of 53.4 0C to 75 0C and 74 0C to 95 0C
respectively. For the temperature transition from 55 0C to 95 0C the surface will be coated
with both set of liquid crystals. Because both set of liquid crystals have to different
response time the CCD camera will be used to capture of the heated surface for the two
separate temperature displayed. The other two temperature transition 55 0C to 72 0C and
72 0C to 95 0C liquid crystal of temperature range 53.4 0C to 75 0C and 74 0C to 95 0C
will be used respectively. The image would be compared to the standard commercial
chart for the actual temperature measurement.
In the future liquid crystal would be used to display the steady state temperature
profile alone the channel. A camera will then be used to capture the image displayed by
the Liquid crystal. These images will then compared to the industrial chart to assign
values to the different colors displayed by Liquid crystal. The temperature contours
obtained from the Liquid crystal were then compared to those achieved from ANSYS
simulation to ensure that the predictions from the simulations were valid
The only experiment that was performed for the thesis was for the transient
temperature response of the system. The temperatures were plotted against the simulated
results for comparison (see Figure 6.7). The experimental results showed a time lag
which resulting from the thermal capacitance of the heaters that was neglected in the
99
numerical simulation. Experiment for the steady state temperature contour using liquid
crystals will be carried out at a later time.
6.7 EXPERIMENTAL SET UP AND RESULTS
CCD Camera
Computer
Thermocouple
readout
Control panel with
three controllers
Test
Device
Power Supply
Figure 6.6: A schematic diagram of the experimental setup to test for the transient
response and capture the steady state temperature contour of the PCR
device.
100
Step Res pons e
370
360
368 C Zone
345 C Zone
328 C Zone
Temperature (K)
350
340
330
320
310
300
0
100
200
300
400
500
600
Time (s ec )
Figure 6.7: The comparison between simulated and experimental transient temperature
response
101
CHAPTER 7: DISCUSSIONS AND CONCLUSION
It was showed that it was possible to fabricate a PCR device from polycarbonate,
which produced amplification comparable to the conventional laboratory-scale device.
An analytical tool was developed to predict the temperature distribution for this specific
continuous flow PCR device configuration. By using air pockets it confirm by the
simulation that it was possible to achieve a relatively uniform temperature zone of
tolerances ± 5 0C the design value for an average fluid velocity of 2mm/s. This
simulation tool showed that the thinner the device the easer it is to obtained the desired
temperature profile. The simulation tool was used for parametric studies and prediction of
the device performance for increasing fluid velocity.
Different materials were investigated for their workable temperature range and
formability and polycarbonate was selected based on its ability to be used for mass
production and working temperatures of up to 148 0C. The polymer chips have better
mechanical properties than the silicon and glass chips and can be mass-produced.
The heating and cooling rates of most microfabricated PCR devices depends on
their thermal capacitance, which dictate the PCR reaction time. Kopp in (19980 showed
that continuous flow PCR devices were independent of their thermal capacitance because
they were used at steady state temperatures. The thermal cycle for a continuous flows
PCR device depend on the device configuration, the thermal capacitance of the fluid and
velocity.
Since high speed processing is a necessity for clinically useful devices a series of
finite element simulations using ANSYS/CFD-FLOTRAN (vers. 5.7, ANSYS, Inc.,
Canonsburg, PA) was performed to evaluate the effects of increasing fluid velocity. It
102
was observed that as the speed increases the fluid spends more time in transition than in
the temperature zones. As the fluid velocity increases another trend was observed, the
entire fluid stream was converging on an average temperature of 74 0C. For a fluid
velocity of 5 mm/s, no heat input was needed to achieve temperature of 60 0C in the
annealing (55 0C) zone. This meant that at fluid velocity of 5 mm/s and higher it was not
possible to control the temperature for annealing, because there would be no control over
the thermal power input. The maximum recommended mean velocity for this particular
device configuration is 4 mm/s with maximum velocity at the center of channel to be 7.5
mm/s, which result in a reaction time of 15 sec per cycle. At this speed it was possible to
maintain temperature within ± 5 0C of design value in each zone. The comparison of the
temperature along the wall of the channel and its center shows a maximum difference of
1.3 0C. This is the maximum lateral temperature variation in each temperature zone for an
average fluid velocity of 4mm/s.
Simulation results showed that single-sided heating produced an adequate
temperature profile, but that it has a thermal transition length approximately twice that of
double side hearting. Operating the PCR device at average velocities higher than 2 mm/s
would increase the magnitude of the thermal power requirement. Leading to an increase
in the temperature beneath the heaters, which may get too close to the polycarbonate
glass transition temperature. The main disadvantage of single-sided heating is that as the
velocity increase the fluid will always be in transition rather than reaching a steady state
temperature. Two-sided heating is recommended for velocities greater than 2 mm/s for
this device configuration.
103
By using X-Ray LIGA to fabricate the mold inserts straight and smooth sidewalls
were achieved. These mold inserts were used to emboss multiple devices. The mold
inserts were fabricated of nickel and steel they are relatively durable, which may permit
massproduction of the devices. SEM photos of the microchannels produced by X-Ray
LIGA mold insert shows that the quality is much better than that produced by UV LIGA
mold insert.
Based on the similarity of the experimental and analytical transient temperatures
response, the assumptions made during the system dynamics simulation were validated.
The controllers and heaters that were chosen to maintain the temperatures in the zones
were appropriate for this application. This device can complete a PCR reaction as fast as
15sec/cycle at volume of 0.45µl / cycle.
7.1 RECOMMENDATIONS AND FUTURE WORKS
The heating element should be deposited directly on the polycarbonate this will
help to ensure that uniform heating is achieved. It will also reduce the effect of the
thermal capacitance of the heaters. Depositing the heating element directly on the
polycarbonate cover plate will increase the accuracy of the approximation of onedimensional heat flow. Improvement in the present molding process so that thinner
polycarbonate can be hot embossed would also help in achieving one-dimensional, heat
flow. The device should be tested for speed of 4mm/s to determine what amplification
would be achieved at this velocity.
To improve the speed of the device, the first stem would be to use a smaller
footprint to construct the device. This will shorten the channel length, resulting in a
104
shorten cycle time. Development of fluidic interconnects for the attachment of the PCR to
the LDR device and simulates their temperature response.
105
REFERENCES
ANSYS 5.4, CDF FLOTRAN Analysis Guide, Canonsburg, PA: ANSYS, Inc., 1997.
ANSYS 5.4, Thermal Analysis Guide, Canonsburg, PA: ANSYS, Inc., 1997.
Bejat, y.d., “Micro-Chip Design, numerical Simulation and Micro-PIV Diagnostics for
DNA Assays”., Master’s Thesis, Louisiana State University, Baton Rouge, LA, 2001.
Burmeister, Louis. C, Convective Heat transfer, 2nd ed.,Third Avenue, New York, NY:
John Wiley and Sons, Inc.,pp. 143., 1993
Burns, M.A., Johnson, B.N., Brahmasandra, S.N., Handique, K., Websterx, J.R.,
Krishnan, M., Sammarco, T.S., Man, P.M., Jones, D., Heldsinger, D., Mastrangelo C.H.,
and Burke, D.T., "An integrated nanoliter DNA analysis device," Science, vol. 282, no.
5388, pp. 484-487, 1998.
Burns, M.A., Mastrangelo, C.H., Sammarco, T.S., Man, F.P., Webster, J.R., Johnsons,
B.N., Foerster, Jones, B., Fields, D.Y., Kaiser, A.R., and Burke, D.T., "Microfabricated
structures for integrated DNA analysis," Proc. Natl. Acad. Sci. USA, vol. 93, no. 11, pp.
5556-5561, 1996.
Cheng, J., Shoffner, M.A., Hvichia, G.E., Kricka, L.J., and Wilding, P., Nucleic Acids
Research, vol. 24, p. 380, 1996.
Desta, Y.M., Aigeldinger, G., Zanca, K.J., Coane, P., Gottert, J., Murphy, M.C.,
“Fabrication of graphite masks for deep and ultra-deep X-ray lithography”, in Materials
and Devices Characterization in Micromachining III, eds. Y. Vladimirsky, P. J. Coane,
Society of Photo-optical Instrumentation Engineers (SPIE), v 4175, pp 122-130, 2000.
Dryden, H. L., Murnaghan, F. D., and Bateman, H., Hydrodynamics, Bull. No. 84,
pp.197-201. Comm. Hydrodyn., Div. Phy. Sci., Res. Counc., Washington, Dc., 1932:
reprinted by Dover, New York, 1956.
Kundu, Pijush. K., Fluid Mechanics, San Diego, California: Academic Press, Inc., 1990,
pp, 77-106.
Ferrance, J. P., Giordano, B., and Landers, J. P., “Toward Effective PCR-Based
Amplification of DNA on Microfabrication Chips” Methods in Molecular Biology, vol.
163, pp. 1-14,2000.
106
Franklin, G.F., Powell, J.D., Emani-Naeini, A., “Feedback Control of Dynamics Systems”
3rd ed., Addison-Wesley Publishing, Inc, Natick, MA., 1994
Friedman, N.A., and Meldrum, D.R., “Capillary tube resistive thermal cycling,” Anal.
Chem., vol. 70, no. 14, pp. 2997-3002, 1998.
Giordano, B.C., Ferrance, J., Swedberg, S., Huhmer, A.F.R., and Landers, J.P.,
“Polymerase chain reaction in polymeric microchips: DNA amplification in less than 240
seconds,” Anal. Biochem.,. vol. 291, no. 1, p. 124-132, 2001.
http://www.matweb.com, January, 20, 2002.
http://www.accessexcellence.org, November, 14, 2002
Henry, A.C., McCarley, R.L., “Selective Deposition of Metal on Plastics Used in the
Construction of Microanalytical Devices: Photo-direction Formation of Metal Features
on PMMA”., J. Phys. Chem. B, vol. 105, pp. 8755-8761, 2001.
Hidenori Nagai, Yuji Murakami, Yasutaka Morita, Kenji Yokoyama, and Eiichi Tamiya,
“Development of A Microchamber Array for Picoliter PCR” Anal. Chem., vol. 73, no. 5,
pp. 1043-1047, 2001.
Huhmer, A.F.R., and Landers, J.P., Anal. Chem,. vol. 72, p. 5507, 2000.
Innis, M.A. Myambo, K.B., Gelfand, D.H. and Brow, M.A.D. “DNA sequencing with
Thermus aquaticus DNA polymerase and direct sequencing of polymerase chain reactionamplified DNA,” Proc. Natl. Acad. Sci. USA , vol. 85, no. 24, pp. 9436-9440. 1988.
Ivonne Schneegass, Reiner Brautigam and Johann Michael Kohler, “Miniaturized flowthrough PCR with different template types in a silicon chip thermocycler”
Jing Cheng, Mann A. Schoffner, Georgi E. Hvichia, Larry J. Kricka and Peter Wilding,
“Chip PCR. II. Investigation of different PCR amplification systems in microfabricated
silicon-glass chips” Nucleic Acids Research, vol. 24, no. 2, pp. 380-385, 1995.
Kakac, S., Yener, Y., Heat Conduction 3rd ed., Vermont Avenue, N.W., Washington, Dc.,
Taylor and Francis. 1993, pp. 34-36
Kalinina, O., Lebedeva, I., Brown, J., and Silver, J., “Nanaliter scale PCR with TaqMan
detection”, Nucleic Acids Research, vol. 25, no. 10, pp 1999-2004, 1997.
Karnopp, D.C., Rosenberg, R.C., “Introduction To Physical System Dynamics”, McGraw
Hill, Inc., USA (2000).
107
Khandurina, J., Mcknight, T.E., Jacobson, S.C., Waters, L.C., Foote, R. S., and Ramsey,
J. M., “Analytical Chemistry”, vol.72, no.13, pp2995-3000, 2000.
Kopp, M.U., DeMello, A.J., and Manz, A., “Chemical amplification: continuous-flow
PCR on a chip,” Science, vol. 280, no. 5366, pp. 1046-1048, 1998.
Kovacs, G.T.A., “Micromachined Transducers Sourceboob”, McGraw Hill, Inc., New
York, NY, USA (1998).
Lagally, E.T., Medintz, I., and Mathies, R.A., “Single-molecule DNA amplification and
analysis in an integrated microfluidic device,” Anal. Chem., vol. 73, no. 3, pp. 565-570,
2001.
Lagally, E. T., Simpson, P. C., Mathies, R. A., “Monololithic integrated microfluidic
DNA amplification and capillary electrolysis analysis system” Sensors and Actuators B,
vol. 63, pp. 138-146, 2000.
Lin, K.L., Ho, J.K., “Electrodeposited Ni-Cr and Ni-Cr-P alloys”., J.Electrochem. Soc.,
vol.139, no. 5, pp. 1305-1310, 1992
Lin, K.L., Hsu, C.J., Hsu, I.M., Chang, J.T., “Electrolating of Ni-Cr on Steel with Pulse
Plating”., Journal of material Engineering and Preforamance., vol. 1, no.3, pp. 359-361,
1992.
Liu, X., Mitchell. M. W., Bejat, Y., Ford M.S., Nikitopoulos, D .E., Soper, S .A.,
Murphy , M. C., “A molded polycarbonate continuous flow polymerase chain reaction
device” To be submitted to J. MEMS
Madou, M., “Fundamentals Of Microfabrication”, CRC Press, Inc., Boca Raton, FL,
USA 1997.
Mann A. Schoffer, Jing Cheng, Georgi E. Hvichia, Larry J. Kricka and Peter Wilding,
“Chip PCR I. Surface passivation of microfabricated silicon-glass chips for PCR” Nucleic
Acids Research, vol. 24, no. 2, pp. 375-379, 1996.
Mullis, K.B. Ferre, F. F., and Gibbs. R.A. The Polymerase Chain Reaction, Birkhauser,
Boston, 1994.
Oda R. P., Straubauch, M. A., A. Huhmer, F.R., Borson, N., Jurrens, S. R., Craighead,
J., Wettstein, P. J., Eckloff, B., Kline, B. , and Landers, J. P., “Infrared-Mediated
Thermocycling for Ultrafast Polymerase Chain Reaction Amplification of DNA” Anal.
Chem., vol. 70, no. 20, pp. 4361-4368, 1998.
108
Pan, L.W., Lin L., (2001), “Batch transfer of LIGA microstructures by selective
electroplating and bonding”, J. Microelectromech. Systems, v 10, n 1, pp 25-32.
Saiki, R.K. Gelfand, D.H. Stoffel, S. Sharf, S.J. Higuchi, S.J., Horn, R. G.T. Mullis, K.B.
and Erlich, H.A. “Primer-directed enzymic amplification of DNA with a thermostable
DNA polymerase,” Science, vol. 239, p. 487, 1988.
Schlesinger, M., Paunovic, M., “ Modern Electroplating” 4th ed., John Wiley & Sons,
Inc, Third Avenue, New York, NY.,2000
Schneegass, R.B., and Kohler, J.M. “Miniaturized flow-through PCR with different
template types in a silicon chip thermocycler,” Lab on a Chip, vol. 1, no. 1, pp. 42-49,
2001.
Stefanov, P.; Stoychev, D.; Stoycheva, M.; Marinova, T. Materials Chemistry and
Physics 2000, 65, 212-215.
Stephen P. Naber. “ Molecular Pathology: Diagnosis of Infectious Disease” 331: 12121215, November 3, 1994, and “Molecular Pathology: detection of Neoplasia The New
England Journal of Medicine.” 331: 1508-1510, December 1, 1994.
Soper, S.A. Ford S.M., Qi, S. McCarley, R.L. Kelly, K. and Murphy, M.C. “Polymeric
microelectromechanical systems,” Anal. Chem., vol. 72, no. 19, pp. 642A-651A, 2000.
Mills, A.F., Heat Transfer, 2nd ed., Upper Saddle River, NJ: Prentice Hall, Inc., 1999, pp.
277-279.
U.S., Congress. Genetic Witness: Forensic Uses of DNA Test. U.S Government Printing
Office OTA-BA-438 July 1990.
Waters, L.C , Jacobson, S.C., Kroutchinina, N., Khandurina, J., Foote, R.S., and Ramsey,
J. M., “Multiple Sample PCR Amplification and Electrophoretic Analysis on a
Microchip”, Anal. Chem., vol.70, pp5172-5176, 1998.
Waters, L.C , Jacobson, S.C., Kroutchinina, N., Khandurina, J., Foote, R.S., and Ramsey,
J. M., “Microchip Device for Cell Lysis, Multiplex PCR Amplification, and
Electrophoretic Sizing”, Anal. Chem., vol.70, pp158-162, 1996.
West, A. C., Matlosz, M., Landolt, D., “Primary current distribution in Hull Cell and
related trapezoidal geometries”, Journal of Applied Electrochemistry, vol. 22, no. 31, pp.
301-303, 1991
109
Wittwer, C.T. Fillmore, G.C. and Hillyard, D.R. D. R., “Automated polymerase chain
reaction in capillary tubes with hot air,” Nucleic Acids Res, vol. 17, no. 11, pp. 43534358, 1989.
Wittwer, C.T. Fillmore, G.C., and Garling, D.J. “Minimizing the time required for DNA
amplification by efficient heat transfer to small samples,” Anal. Biochem. , vol. 186, no.
2, pp. 328-331, 1990.
Woolley, A.T., Hadley, D., Landre, P., DeMello, A.J., Mathies, R.A., and Northrup,
M.A., “Functional integration of PCR amplification and capillary electrophoresis in a
microfabricated DNA analysis device,” Anal. Chem, vol. 68, no. 23, pp. 4081-4086,
1996.
110
APPENDIX A: A SCHEMATIC OF THE FABRICATION PROCEDURES USED TO
MAKE THE GRAPHITE X-RAY MASK
Spin coating of
40 µm SU 8-25
Graphite
UV Light
Optical Cr
Mask
Development
of SU 8-25
15µm Gold Electroplating
At 4µm per hour & 65oC
X-Ray Mask
111
APPENDIX B: SCHEMATIC OF THE FABRICATION PROCEDURES USED TO
MAKE THE TWO MOLD INSERTS USING THE X-RAY MASK
Resist PMMA
Steel Plating
Base
X-Ray Radiation
X-Ray mask
Exposed Resist
Developed Resist
Electrodeposition of Nickel
which was Lapped and polished
Flood exposure and
Resist Development to
produced Mold Insert
112
APPENDIX C: A COMPLETE DIMENSION DRAWING OF THE TEST DEVICE
Optical Mask
pattern
Cover plate with
air pockets
113
VITA
Michael Mitchell was born on May 12, 1974, in Clarendon Jamaica. He received
his Bachelor of Science degree in mechanical engineering in December of 2000 from
Louisiana State University, Baton Rouge La , USA.
He joined the graduate program in mechanical engineering at Louisiana State
University in 2000 as a member of the Microsystems Engineering Team with the
Mechanical Engineering Department and started working on this project in June 1999. He
expects to receive the degree of Master of Science in Mechanical Engineering in
December 2002.
114